From d64b788e56400ac52755d03fe81ef4ae06250101 Mon Sep 17 00:00:00 2001 From: wassname <1103714+wassname@users.noreply.github.com> Date: Fri, 18 Sep 2026 18:39:34 +0800 Subject: [PATCH] Preregister Gemini Flash rubric pilot Signed-off-by: PI[gpt-5.6-terra] --- docs/RESEARCH_JOURNAL.md | 65 + .../wvs_api/08_gemini_flash_rubric_pilot.sh | 3 + scripts/wvs_gemini_flash_rubric_pilot.py | 791 + ...60918_gemini_flash_high_reasoning_smoke.md | 128 + ..._gemini_flash_high_reasoning_smoke_2048.md | 92 + .../budget.json | 10 + .../endpoint_catalog.json | 31114 ++++++++++++++++ .../manifest.json | 673 + .../metadata_capture_smoke.json | 33 + .../metadata_capture_smoke.jsonl | 11 + .../paid_smoke.json | 273 + .../paid_smoke.jsonl | 8 + .../paid_smoke_2048.json | 154 + .../paid_smoke_2048.jsonl | 6 + .../paid_smoke_2048_run.log | 25 + .../paid_smoke_run.log | 25 + .../quantization_audit.json | 1317 + .../request_attempts.jsonl | 3 + .../rescue_seed_capture_smoke.jsonl | 8 + .../route_mismatch_capture_smoke.json | 40 + src/moralmaps/read_api.py | 142 +- 21 files changed, 34894 insertions(+), 27 deletions(-) create mode 100755 scripts/wvs_api/08_gemini_flash_rubric_pilot.sh create mode 100644 scripts/wvs_gemini_flash_rubric_pilot.py create mode 100644 slop/audits/20260918_gemini_flash_high_reasoning_smoke.md create mode 100644 slop/audits/20260918_gemini_flash_high_reasoning_smoke_2048.md create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/budget.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/manifest.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.jsonl create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048_run.log create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_run.log create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/quantization_audit.json create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/request_attempts.jsonl create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/rescue_seed_capture_smoke.jsonl create mode 100644 slop/research/wvs/20260918_gemini_flash_rubric_pilot/route_mismatch_capture_smoke.json diff --git a/docs/RESEARCH_JOURNAL.md b/docs/RESEARCH_JOURNAL.md index aa1ad0a..560f581 100644 --- a/docs/RESEARCH_JOURNAL.md +++ b/docs/RESEARCH_JOURNAL.md @@ -1026,3 +1026,68 @@ The ledger's 146 accepted phases sum to 28,157 prompt tokens, 78,117 completion My read: this is very probably a valid exact-protocol mechanics diagnostic and resolves the task 1622 wrapper/config regression. The two rescues are explicit and billed rather than hidden, but their upstream empty-message cause remains unknown. An independent raw-answer review found 11 of 12 Homosexuality pole replies rate every mutually exclusive option identically, producing an expected score 5.57 near the 5.5 midpoint. The parse-completeness result therefore does not validate the self-expression coordinate for priority sequencing. Priority dispatch is paused pending an explicit content-quality metric and its cheapest diagnostic. + +## 2026-09-18 -- Gemini Flash reasoning and rating-rubric pilot design + +This entry preregisters a provider-locked pilot to separate reasoning-depth effects from sensitivity to the rating rubric. + +The saved model catalog and the endpoint snapshot at `slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json` identify five compatible full Gemini Flash releases: `google/gemini-3-flash-preview`, `google/gemini-3.5-flash`, `google/gemini-3.6-flash`, `google/gemini-3.7-flash`, and `google/gemini-3.8-flash`. Gemini 3 Flash has only a preview entry in the saved catalog, so it is retained as that release. Batch aliases, Flash Lite, and image variants are excluded. Gemini 2.5 Flash is excluded because its Google AI Studio endpoint advertises `reasoning` but not `reasoning_effort`, so the required minimum-versus-high comparison is not established. The five selected standard Google AI Studio endpoints advertise both `reasoning_effort` and structured output. Their advertised quantization is `unknown`. + +Each model has four cells: normal human rubric at minimum and high reasoning, and reversed rating rubric at minimum and high reasoning. Minimum is `minimal` for Gemini 3, 3.5, and 3.6 Flash, and `low` for Gemini 3.7 and 3.8 Flash. Each cell uses the same twelve WVS items and options, six samples per item, paired deterministic seeds, and three canonical plus three reversed option orders for binary items. The reversed rubric says that one is strong endorsement and five is strong rejection; analysis applies `6 - rating`, but retains the raw reversed cell separately and does not merge it into the primary human-comparable result. There is no z-scaling. + +The provider policy is `only=["google-ai-studio"]`, `allow_fallbacks=false`, and `require_parameters=true`. A complete pilot is seventy-two requests per cell, two hundred eighty-eight per model, and fourteen hundred forty paid panel requests, plus one paid smoke request. The runtime reserves 2048 input tokens and 1024 output tokens for each initial request at the saved standard-endpoint prices. This bounds the panel at USD 9.363456 and the smoke at USD 0.004096, leaving USD 0.632448 below the USD 10 hard stop. Each request permits at most three attempts; a failed attempt is retained and charged at its conservative bound. A later run reuses each parse-valid saved request and only calls missing samples. Rescues and retries can therefore stop the pilot before completion rather than crossing the cap. + +The quantization audit covers all ninety-seven exact model IDs in the canonical v1 cache and all 15,772 matching completed request phases in the saved ledger. No response contains an explicit exact quantization. Of these phases, 5,196 have a saved provider that joins to a currently listed endpoint with a known quantization, but that join is ambiguous because the endpoint snapshot was fetched after the requests and is not historical route evidence. The remaining 10,576 have unknown quantization under the current join. Source: `slop/research/wvs/20260918_gemini_flash_rubric_pilot/quantization_audit.json` and its dated endpoint snapshot. My read: quantization-stratified variance is not identifiable from the saved results; provider and current endpoint metadata must not be relabelled as the historical quantization. + +Predictions recorded before requests: + +- If reasoning depth is a material source of coordinate variation, the paired high-minus-minimum shift should repeat in direction across releases within each rubric. +- If rating-scale wording is a material source, the reverse-transformed reversed-rubric cell should differ from the normal-rubric cell under the same reasoning setting. +- If a release trend is robust, its direction should be similar in all four cells. A trend that changes sign across rubric or reasoning cells is evidence of measurement sensitivity rather than a stable family trajectory. +- Under a rubric-invariant readout, normal and reverse-transformed cells should agree within their paired sampling uncertainty. + +My read: the crossed design is likely more informative than increasing repeats under one prompt, because it tests two named measurement choices while holding provider, items, options, and seed schedules fixed. A remaining alternative is genuine model variation within the Flash series; the four-cell agreement pattern is what separates that from prompt sensitivity. -- PI[gpt-5.6-terra] + +The paid run begins only after metadata capture is durable and a one-request smoke record shows the actual route and response fields. + +## 2026-09-18 -- Gemini Flash high-reasoning paid smoke + +This entry records the provider-locked smoke before the crossed Gemini Flash pilot. + +| phase | finish | prompt tokens | completion tokens | reasoning tokens | cost, USD | parse result | +| --- | --- | ---: | ---: | ---: | ---: | --- | +| initial | length | 166 | 1,009 | 979 | 0.0031100 | truncated JSON | +| rescue | stop | 525 | 662 | 602 | 0.0022485 | valid JSON | +| total | | 691 | 1,671 | 1,581 | 0.0053585 | one valid sample | + +The raw records are `slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl`; the checked summary is `paid_smoke.json` in the same directory. Both phases report requested and response model `google/gemini-3-flash-preview`, selected provider `Google AI Studio`, and selected release slug `google/gemini-3-flash-preview-20251217`. The saved standard endpoint advertises quantization as `unknown`. Local conservative spend and provider-reported spend both equal USD 0.0053585, with no failed phase, no held reservation, and the same amount settled to the repository-wide budget ledger. + +My read: high reasoning can exhaust the initial completion budget before returning the rating object, because this initial phase spent 979 of 1,009 completion tokens on reasoning and stopped for length. The rescue path recovered this sample, but one sample does not estimate how often rescue will be needed. Repeated rescues would make the preregistered initial-request cost bound optimistic for completion, while the USD 10 runtime stop would still end the pilot early rather than overspend. -- PI[gpt-5.6-terra] + +The full pilot remains unqueued until this smoke is reviewed. + +## 2026-09-18 -- Gemini Flash protocol amendment after the high-reasoning smoke + +This entry records the protocol change made before any panel run. + +The first high-reasoning smoke stopped for length after using 979 reasoning tokens inside 1,009 completion tokens, then required a rescue. Source: `slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl:3-5`. The initial USD 9.363456 panel estimate assumed 1,024 output tokens for every cell and did not bound rescue phases. It is therefore not a completion-safe estimate for the observed high-reasoning path. + +The amended protocol holds `max_tokens=2048` fixed in all four cells, so output budget is not a second treatment that differs with reasoning condition. N remains six per item and cell across the same five releases. Each cell's 2,048-token setting is part of its manifest row and protocol identity. The request reservation holds input at 2,048 tokens for both initial and rescue phases; the observed initial and rescue prompts used 166 and 525 prompt tokens, so increasing the input reserve with the output limit had no measured basis. The regenerated manifest bounds the 1,440 initial panel requests at USD 16.220160. One revised smoke is bounded at USD 0.007168, leaving USD 3.772672 below the owner-authorized USD 20 stage stop before observed spend. Sources: `slop/research/wvs/20260918_gemini_flash_rubric_pilot/manifest.json` and the standard-endpoint prices in `endpoint_catalog.json`. + +The rescue now receives the same paired seed as its initial request. Every paid response is route-validated before the caller can accept it. A route mismatch is charged using returned usage, saved with the full response as `request_attempt_route_invalid`, and fails the sample rather than retrying another route. The offline smoke verifies seed propagation, route-mismatch settlement and persistence, bounded transient retry, request reuse, and the reversed-rating transform. Source: `slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.json` and its JSONL records. + +My read: fixing the token budget across cells removes a clear condition-specific confound while preserving reasoning effort and rubric direction as the two intended treatments. The larger initial-request bound is an estimate, while the runtime stage stop remains the spending limit. A new provider-locked high-reasoning smoke must return parse-valid JSON without rescue before the full pilot can be considered for dispatch. -- PI[gpt-5.6-terra] + +The amended full pilot remains unqueued pending the revised paid smoke and review. + +## 2026-09-18 -- Revised Gemini Flash high-reasoning smoke + +This entry records the paid smoke of the amended fixed-output-budget protocol. + +The `normal_high` cell used `google/gemini-3-flash-preview`, high reasoning, `max_tokens=2048`, and the same first-item seed as its full-panel manifest. The initial request returned a complete ten-key rating object with `finish_reason=stop`; no rescue was dispatched. It used 166 prompt tokens and 1,535 completion tokens, of which 1,494 were reasoning tokens, for 1,701 total tokens and USD 0.004688. The selected route was Google AI Studio release `google/gemini-3-flash-preview-20251217`; the standard endpoint's advertised quantization field was present as `unknown`. Source: `slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl` and checked summary `paid_smoke_2048.json`. + +The local pilot ledger now includes both smoke protocols and reports USD 0.0100465 as provider-reported and conservative spend, zero held reservation, and zero failed phases. The revised smoke's distinct repository-wide settlement is USD 0.004688 with no matching held reservation. Replaying the smoke command reused the completed protocol and added no request-attempt record. + +My read: this sample supports the operational requirement that the fixed 2,048-token output budget can return high-reasoning ratings without a rescue. It does not estimate truncation frequency across items or releases. The fixed budget removes the known condition-specific budget difference, while route validation, seeded rescue, the USD 20 stage stop, and the append-only records bound the remaining operational risks. -- PI[gpt-5.6-terra] + +The full pilot remains unqueued until the revised smoke is inspected. diff --git a/scripts/wvs_api/08_gemini_flash_rubric_pilot.sh b/scripts/wvs_api/08_gemini_flash_rubric_pilot.sh new file mode 100755 index 0000000..f1dd7ee --- /dev/null +++ b/scripts/wvs_api/08_gemini_flash_rubric_pilot.sh @@ -0,0 +1,3 @@ +#!/bin/sh +set -eu +exec uv run --offline --with 'datasets>=4.0,<5' python scripts/wvs_gemini_flash_rubric_pilot.py --run diff --git a/scripts/wvs_gemini_flash_rubric_pilot.py b/scripts/wvs_gemini_flash_rubric_pilot.py new file mode 100644 index 0000000..a3f8fc5 --- /dev/null +++ b/scripts/wvs_gemini_flash_rubric_pilot.py @@ -0,0 +1,791 @@ +#!/usr/bin/env python3 +"""Run the provider-locked Gemini Flash reasoning and rating-rubric pilot. -- PI[gpt-5.6-terra]""" +from __future__ import annotations + +import argparse +import asyncio +import fcntl +import hashlib +import json +import os +from collections import Counter +from datetime import UTC, datetime +from decimal import Decimal +from pathlib import Path + +import httpx +import numpy as np + +from moralmaps.iw_axes import X_AXIS, Y_AXIS, resolve_items +from moralmaps.read_api import ( + openrouter_request_with_metadata_once, + rated_protocol_identity, + read_items_rated, +) +from openrouter_wrapper.retry import is_retryable_error +from wvs_map import _sample_only_coord_se, load_wvs_all, model_coord_ci +from wvs_score_all_options_refresh import STATE as GLOBAL_BUDGET_STATE +from wvs_score_all_options_refresh import reserve, settle_external_reservation + +EVAL_VERSION = "wvs-gemini-flash-rubric-v1" +MODELS = ( + "google/gemini-3-flash-preview", + "google/gemini-3.5-flash", + "google/gemini-3.6-flash", + "google/gemini-3.7-flash", + "google/gemini-3.8-flash", +) +MINIMUM_REASONING = { + "google/gemini-3-flash-preview": "minimal", + "google/gemini-3.5-flash": "minimal", + "google/gemini-3.6-flash": "minimal", + "google/gemini-3.7-flash": "low", + "google/gemini-3.8-flash": "low", +} +PROVIDER = {"only": ["google-ai-studio"], "allow_fallbacks": False, "require_parameters": True} +CELLS = ( + ("normal_minimum", "normal", "minimum"), + ("normal_high", "normal", "high"), + ("reversed_minimum", "reversed", "minimum"), + ("reversed_high", "reversed", "high"), +) +N_SAMPLES = 6 +MAX_TOKENS = 2048 +OUT = Path("slop/research/wvs/20260918_gemini_flash_rubric_pilot") +ENDPOINT_CATALOG = OUT / "endpoint_catalog.json" +MANIFEST = OUT / "manifest.json" +QUANTIZATION_AUDIT = OUT / "quantization_audit.json" +RESULTS = OUT / "results.json" +STATE = OUT / "budget.json" +LOCK = OUT / "budget.lock" +EXISTING_LEDGER = Path("slop/research/wvs/20260916_openrouter/wvs_iw_requests.jsonl") +CANONICAL_CACHE = Path("slop/research/wvs/20260916_openrouter/wvs_iw_rated.json") +MODEL_CATALOG = Path("slop/research/wvs/20260917_openrouter_models.json") +REQUEST_ATTEMPTS = OUT / "request_attempts.jsonl" +PILOT_CAP_USD = Decimal("20") +MAX_ATTEMPTS = 3 +TRANSPORT_TIMEOUT = 120 +REQUEST_TIMEOUT = 400 +GLOBAL_RESERVATION = "pilot/gemini-flash-rubric" +SMOKE_RESERVATION = "pilot/gemini-flash-rubric-smoke-2048" +REVISED_SMOKE_RECORDS = OUT / "paid_smoke_2048.jsonl" +REVISED_SMOKE_SUMMARY = OUT / "paid_smoke_2048.json" + + +def atomic_json(path: Path, value: object) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + temp = path.with_suffix(path.suffix + ".tmp") + temp.write_text(json.dumps(value, indent=2, sort_keys=True) + "\n") + temp.replace(path) + + +def reservation_id(prefix: str) -> str: + return f"{prefix}/{datetime.now(UTC).strftime('%Y%m%dT%H%M%S.%fZ')}" + + +def rated_items() -> tuple[list[dict], dict]: + resolved = resolve_items(load_wvs_all()) + items, seen = [], set() + for axis in (X_AXIS, Y_AXIS): + for item in resolved[axis]: + if item["suffix"] in seen: + continue + seen.add(item["suffix"]) + items.append({"id": item["suffix"], "question": item["rec"]["q"], + "options": item["rec"]["opts"], "n": item["n"]}) + return items, resolved + + +def model_catalog() -> dict[str, dict]: + return {row["id"]: row for row in json.loads(MODEL_CATALOG.read_text())["data"]} + + +def endpoint_catalog() -> dict[str, dict]: + return {row["id"]: row for row in json.loads(ENDPOINT_CATALOG.read_text())["models"]} + + +def standard_endpoint(model: str) -> dict: + matches = [row for row in endpoint_catalog()[model]["endpoints"] + if row.get("tag") == "google-ai-studio"] + if len(matches) != 1: + raise RuntimeError(f"expected one standard google-ai-studio endpoint for {model}, got {len(matches)}") + return matches[0] + + +def validate_route(response: dict, model: str) -> dict: + endpoint = standard_endpoint(model) + if "quantization" not in endpoint: + raise RuntimeError(f"standard endpoint omitted advertised quantization: {model}") + prefix = f"{endpoint['provider_name']} | " + if not endpoint["name"].startswith(prefix): + raise RuntimeError(f"cannot recover exact release slug from endpoint name: {endpoint['name']}") + exact_release_slug = endpoint["name"].removeprefix(prefix) + metadata = response.get("openrouter_metadata") + if not isinstance(metadata, dict): + raise RuntimeError("response omitted openrouter_metadata") + available = metadata.get("endpoints", {}).get("available", []) + selected = [row for row in available if row.get("selected") is True] + if len(selected) != 1: + raise RuntimeError(f"expected one selected metadata endpoint, got {len(selected)}") + route = selected[0] + expected = { + "requested_model": model, + "response_model": model, + "selected_provider": endpoint["provider_name"], + "selected_release_slug": exact_release_slug, + } + observed = { + "requested_model": metadata.get("requested"), + "response_model": response.get("model"), + "selected_provider": route.get("provider"), + "selected_release_slug": route.get("model"), + } + if observed != expected: + raise RuntimeError(f"route mismatch: expected={expected} observed={observed}") + return { + **observed, + "endpoint_tag": endpoint["tag"], + "advertised_quantization": endpoint["quantization"], + } + + +def deterministic_seeds(model: str, count: int) -> list[int]: + seeds = [] + for sequence in range(count): + digest = hashlib.sha256(f"{EVAL_VERSION}|{model}|paired|{sequence}".encode()).digest() + seeds.append(int.from_bytes(digest[:4], "big") & 0x7FFF_FFFF) + if len(set(seeds)) != len(seeds): + raise RuntimeError(f"seed collision for {model}") + return seeds + + +def reasoning(model: str, level: str) -> dict: + effort = MINIMUM_REASONING[model] if level == "minimum" else "high" + return {"effort": effort} + + +def records_path(model: str, cell: str) -> Path: + return OUT / "records" / model.replace("/", "__") / f"{cell}.jsonl" + + +def explicit_quantizations(value: object) -> list[str]: + found = [] + if isinstance(value, dict): + for key, child in value.items(): + if key.lower() == "quantization" and child not in (None, "unknown"): + found.append(str(child)) + found.extend(explicit_quantizations(child)) + elif isinstance(value, list): + for child in value: + found.extend(explicit_quantizations(child)) + return found + + +def quantization_audit() -> dict: + cache = json.loads(CANONICAL_CACHE.read_text())["completed"] + canonical_models = sorted({entry["model"] for entry in cache.values() + if entry.get("eval_version") == "wvs-score-all-options-v1"}) + endpoints_by_model = endpoint_catalog() + counts = {model: Counter() for model in canonical_models} + providers = {model: Counter() for model in canonical_models} + for line in EXISTING_LEDGER.read_text().splitlines(): + record = json.loads(line) + model = record.get("model") + if model not in counts or record.get("event") != "request_completed": + continue + response = record.get("response") or {} + provider = record.get("provider") or response.get("provider") + providers[model][str(provider)] += 1 + exact = explicit_quantizations(response) + if exact: + counts[model]["exact_response_quantization"] += 1 + continue + if not provider: + counts[model]["unknown_missing_provider"] += 1 + continue + matches = [endpoint for endpoint in endpoints_by_model[model]["endpoints"] + if endpoint.get("provider_name") == provider] + known = {endpoint.get("quantization") for endpoint in matches + if endpoint.get("quantization") not in (None, "unknown")} + if known: + counts[model]["ambiguous_current_snapshot_join"] += 1 + elif matches: + counts[model]["unknown_current_quantization"] += 1 + else: + counts[model]["unknown_no_current_provider_join"] += 1 + rows = [{ + "model": model, + "saved_provider_counts": dict(providers[model]), + "classification_counts": dict(counts[model]), + "current_endpoint_quantizations": sorted({ + str(endpoint.get("quantization")) + for endpoint in endpoints_by_model[model]["endpoints"] + }), + } for model in canonical_models] + exact = sum(row["classification_counts"].get("exact_response_quantization", 0) for row in rows) + ambiguous = sum(row["classification_counts"].get("ambiguous_current_snapshot_join", 0) for row in rows) + unknown = sum(sum(count for key, count in row["classification_counts"].items() + if key.startswith("unknown_")) for row in rows) + total = exact + ambiguous + unknown + return { + "schema": 2, + "source_ledger": str(EXISTING_LEDGER), + "canonical_cache": str(CANONICAL_CACHE), + "endpoint_catalog": str(ENDPOINT_CATALOG), + "endpoint_snapshot_scope": "all exact model IDs in the canonical v1 cache", + "endpoint_snapshot_is_historical_route_evidence": False, + "canonical_models": len(canonical_models), + "completed_request_phases": total, + "classification_counts": {"exact": exact, "ambiguous": ambiguous, "unknown": unknown}, + "rows": rows, + "conclusion": ("No quantization-stratified variance is identifiable. Exact requires an explicit saved " + "response quantization; current endpoint joins are ambiguous rather than historical route evidence."), + } + + +def write_manifest() -> dict: + items, _ = rated_items() + catalog = model_catalog() + rows = [] + for model in MODELS: + endpoint = standard_endpoint(model) + advertised = set(endpoint["supported_parameters"]) + efforts = set(catalog[model]["reasoning"]["supported_efforts"]) + minimum = MINIMUM_REASONING[model] + if not {"reasoning_effort", "structured_outputs", "seed"} <= advertised: + raise RuntimeError(f"endpoint lacks required parameters: {model}") + if not {minimum, "high"} <= efforts: + raise RuntimeError(f"catalog lacks required reasoning efforts: {model}") + seeds = deterministic_seeds(model, len(items) * N_SAMPLES) + cells = [] + for cell, rubric, level in CELLS: + protocol_id = rated_protocol_identity( + model, items, n_samples=N_SAMPLES, temperature=1.0, max_tokens=MAX_TOKENS, + concurrency=1, req_timeout=REQUEST_TIMEOUT, reasoning=reasoning(model, level), + structured_output=True, provider=PROVIDER, eval_version=EVAL_VERSION, + seed_schedule=seeds, rating_rubric=rubric, request_metadata=True, + ) + cells.append({"name": cell, "rating_rubric": rubric, + "reasoning": reasoning(model, level), "max_tokens": MAX_TOKENS, + "protocol_id": protocol_id, "records": str(records_path(model, cell))}) + prompt_price = Decimal(endpoint["pricing"]["prompt"]) * Decimal(1_000_000) + completion_price = Decimal(endpoint["pricing"]["completion"]) * Decimal(1_000_000) + calls = Decimal(len(cells) * len(items) * N_SAMPLES) + bound = calls * (Decimal(2048) * prompt_price + Decimal(MAX_TOKENS) * completion_price) / Decimal(1_000_000) + rows.append({ + "id": model, + "created": catalog[model]["created"], + "minimum_reasoning": minimum, + "paired_seed_schedule": seeds, + "provider_endpoint": endpoint["tag"], + "advertised_quantization": endpoint.get("quantization"), + "input_usd_per_million": str(prompt_price), + "output_usd_per_million": str(completion_price), + "cells": cells, + "calls": int(calls), + "runtime_initial_request_bound_usd": str(bound), + }) + audit = quantization_audit() + panel_bound = sum(Decimal(row["runtime_initial_request_bound_usd"]) for row in rows) + smoke_endpoint = standard_endpoint(MODELS[0]) + smoke_bound = (Decimal(2048) * Decimal(smoke_endpoint["pricing"]["prompt"]) + + Decimal(MAX_TOKENS) * Decimal(smoke_endpoint["pricing"]["completion"])) + payload = { + "schema": 1, + "eval_version": EVAL_VERSION, + "created_utc": datetime.now(UTC).isoformat(), + "models": rows, + "items": len(items), + "samples_per_item_per_cell": N_SAMPLES, + "cells_per_model": len(CELLS), + "panel_calls": sum(row["calls"] for row in rows), + "paid_smoke_calls": 1, + "provider": PROVIDER, + "temperature": 1.0, + "structured_output": True, + "request_metadata": True, + "max_attempts_per_request": MAX_ATTEMPTS, + "transport_timeout_seconds": TRANSPORT_TIMEOUT, + "request_timeout_seconds": REQUEST_TIMEOUT, + "reverse_transform": "6 - rating; raw and transformed samples retained separately", + "merge_reversed_into_primary": False, + "z_scaling": False, + "runtime_input_token_reserve_per_initial_request": 2048, + "panel_initial_request_bound_usd": str(panel_bound), + "paid_smoke_bound_usd": str(smoke_bound), + "panel_plus_smoke_bound_usd": str(panel_bound + smoke_bound), + "headroom_below_hard_stop_usd": str(PILOT_CAP_USD - panel_bound - smoke_bound), + "hard_pilot_stop_usd": str(PILOT_CAP_USD), + "catalog_sha256": hashlib.sha256(MODEL_CATALOG.read_bytes()).hexdigest(), + "endpoint_catalog_sha256": hashlib.sha256(ENDPOINT_CATALOG.read_bytes()).hexdigest(), + "not_published": True, + "quantization_audit": audit["conclusion"], + } + atomic_json(QUANTIZATION_AUDIT, audit) + atomic_json(MANIFEST, payload) + return payload + + +def update_state(update) -> dict: + OUT.mkdir(parents=True, exist_ok=True) + with LOCK.open("w") as lock: + fcntl.flock(lock, fcntl.LOCK_EX) + state = json.loads(STATE.read_text()) if STATE.exists() else { + "schema": 1, + "hard_cap_usd": str(PILOT_CAP_USD), + "provider_reported_spent_usd": "0", + "conservative_spent_usd": "0", + "reserved_usd": "0", + "completed_phases": 0, + "completed_phases_without_provider_cost": 0, + "failed_phases_charged_at_bound": 0, + } + state["hard_cap_usd"] = str(PILOT_CAP_USD) + update(state) + atomic_json(STATE, state) + fcntl.flock(lock, fcntl.LOCK_UN) + return state + + +def request_bound(payload: dict) -> Decimal: + endpoint = standard_endpoint(payload["model"]) + prompt_price = Decimal(endpoint["pricing"]["prompt"]) + completion_price = Decimal(endpoint["pricing"]["completion"]) + input_token_bound = Decimal(2048) + return input_token_bound * prompt_price + Decimal(payload["max_tokens"]) * completion_price + + +def reserve_request(payload: dict) -> Decimal: + bound = request_bound(payload) + def update(state: dict) -> None: + spent = Decimal(state["conservative_spent_usd"]) + held = Decimal(state["reserved_usd"]) + if spent + held + bound > PILOT_CAP_USD: + raise RuntimeError(f"pilot hard stop: spent={spent} held={held} next={bound} cap={PILOT_CAP_USD}") + state["reserved_usd"] = str(held + bound) + update_state(update) + return bound + + +def append_attempt(record: dict) -> None: + record["recorded_at_utc"] = datetime.now(UTC).isoformat() + REQUEST_ATTEMPTS.parent.mkdir(parents=True, exist_ok=True) + with REQUEST_ATTEMPTS.open("a") as file: + file.write(json.dumps(record, sort_keys=True) + "\n") + file.flush() + os.fsync(file.fileno()) + + +def settle_request(bound: Decimal, response: dict | None) -> None: + reported = None if response is None else (response.get("usage") or {}).get("cost") + charged = bound if reported is None else Decimal(str(reported)) + def update(state: dict) -> None: + state["reserved_usd"] = str(Decimal(state["reserved_usd"]) - bound) + state["conservative_spent_usd"] = str(Decimal(state["conservative_spent_usd"]) + charged) + if response is None: + state["failed_phases_charged_at_bound"] += 1 + return + state["completed_phases"] += 1 + if reported is None: + state["completed_phases_without_provider_cost"] += 1 + else: + state["provider_reported_spent_usd"] = str( + Decimal(state["provider_reported_spent_usd"]) + Decimal(str(reported))) + update_state(update) + + +async def budgeted_request(payload: dict) -> dict: + payload_hash = hashlib.sha256(json.dumps(payload, sort_keys=True).encode()).hexdigest() + for attempt in range(1, MAX_ATTEMPTS + 1): + bound = reserve_request(payload) + try: + response = await openrouter_request_with_metadata_once(payload, timeout=TRANSPORT_TIMEOUT) + except Exception as error: + settle_request(bound, None) + response_text = None + status_code = None + if isinstance(error, httpx.HTTPStatusError): + response_text = error.response.text + status_code = error.response.status_code + append_attempt({"event": "request_attempt_failed", "model": payload["model"], + "payload_sha256": payload_hash, "attempt": attempt, + "reserved_bound_usd": str(bound), "error_type": type(error).__name__, + "error": str(error), "error_data": getattr(error, "data", None), + "status_code": status_code, "response_text": response_text}) + if attempt == MAX_ATTEMPTS or not is_retryable_error(error, response_text or ""): + raise + await asyncio.sleep(2 ** (attempt - 1)) + continue + try: + route = validate_route(response, payload["model"]) + except Exception as error: + settle_request(bound, response) + append_attempt({"event": "request_attempt_route_invalid", "model": payload["model"], + "payload_sha256": payload_hash, "attempt": attempt, + "reserved_bound_usd": str(bound), + "provider_reported_cost_usd": (response.get("usage") or {}).get("cost"), + "response_id": response.get("id"), "response": response, + "error_type": type(error).__name__, "error": str(error)}) + raise + settle_request(bound, response) + append_attempt({"event": "request_attempt_completed", "model": payload["model"], + "payload_sha256": payload_hash, "attempt": attempt, + "reserved_bound_usd": str(bound), + "provider_reported_cost_usd": (response.get("usage") or {}).get("cost"), + "response_id": response.get("id"), "route": route, + "openrouter_metadata": response.get("openrouter_metadata")}) + return response + raise AssertionError("unreachable") + + +def cell_result(model: str, cell: dict, items: list[dict], resolved: dict) -> dict | None: + path = Path(cell["records"]) + events = [json.loads(line) for line in path.read_text().splitlines()] if path.exists() else [] + finished = [event for event in events if event.get("event") == "run_finished" + and event.get("protocol_id") == cell["protocol_id"]] + if not finished: + return None + final = finished[-1] + rows = {event["id"]: event for event in events + if event.get("event") == "item_result" and event.get("run_id") == final["run_id"]} + result = { + "name": cell["name"], "protocol_id": cell["protocol_id"], "run_id": final["run_id"], + "records": str(path), "valid_samples": final["valid_samples"], + "failed_samples": final["failed_samples"], "rescued_samples": final["rescued_samples"], + } + complete = (len(rows) == len(items) and final["valid_samples"] == len(items) * N_SAMPLES + and final["failed_samples"] == 0 + and all(row["valid_samples"] == N_SAMPLES for row in rows.values())) + if not complete: + return result | {"status": "incomplete"} + psamples = {item["id"]: np.asarray(rows[item["id"]]["p_samples"]) for item in items} + coords = model_coord_ci(psamples, resolved, np.random.default_rng(0)) + response_se = _sample_only_coord_se(psamples, resolved, np.random.default_rng(1), n_draws=N_SAMPLES) + metadata = [event["response"].get("openrouter_metadata") for event in events + if event.get("event") == "request_completed" and event.get("run_id") == final["run_id"]] + return result | {"status": "complete", "coords": list(coords), + "response_mean_se": list(response_se), + "routing_metadata_records": len([row for row in metadata if row is not None])} + + +def run_cell(model: dict, cell: dict, items: list[dict], resolved: dict) -> dict: + prior = cell_result(model["id"], cell, items, resolved) + if prior is not None and prior["status"] == "complete": + return prior + rows = read_items_rated( + model["id"], items, n_samples=N_SAMPLES, temperature=1.0, max_tokens=cell["max_tokens"], + concurrency=1, req_timeout=REQUEST_TIMEOUT, reasoning=cell["reasoning"], structured_output=True, + records_path=cell["records"], verbose_first=True, provider=PROVIDER, probe_first=True, + eval_version=EVAL_VERSION, identity_eval_version=EVAL_VERSION, + seed_schedule=model["paired_seed_schedule"], rating_rubric=cell["rating_rubric"], + request_metadata=True, reuse_valid_records=True, request_fn=budgeted_request, + ) + if sum(row["valid_samples"] for row in rows) == 0: + raise RuntimeError(f"cell produced no valid samples: {model['id']} {cell['name']}") + result = cell_result(model["id"], cell, items, resolved) + if result is None: + raise RuntimeError(f"cell did not write run_finished: {model['id']} {cell['name']}") + return result + + +def run() -> None: + data = json.loads(MANIFEST.read_text()) + items, resolved = rated_items() + global_reservation = reservation_id(GLOBAL_RESERVATION) + if not reserve({"id": global_reservation, "lane": "google", "reserve_usd": str(PILOT_CAP_USD)}): + raise RuntimeError("global repository cap rejected Gemini pilot reservation") + before = Decimal(update_state(lambda state: state)["conservative_spent_usd"]) + results = {"schema": 1, "eval_version": EVAL_VERSION, "models": [], "not_published": True} + try: + for model in data["models"]: + cells = [] + for cell in model["cells"]: + cells.append(run_cell(model, cell, items, resolved)) + atomic_json(RESULTS, results | {"models": results["models"] + [ + {"id": model["id"], "cells": cells}]}) + results["models"].append({"id": model["id"], "cells": cells}) + atomic_json(RESULTS, results) + finally: + after = Decimal(update_state(lambda state: state)["conservative_spent_usd"]) + settle_external_reservation(global_reservation, after - before) + state = update_state(lambda value: value.update({"finished_utc": datetime.now(UTC).isoformat()})) + if RESULTS.exists(): + saved = json.loads(RESULTS.read_text()) + atomic_json(RESULTS, saved | {"budget": { + "provider_reported_spent_usd": state["provider_reported_spent_usd"], + "conservative_spent_usd": state["conservative_spent_usd"], + "completed_phases_without_provider_cost": state["completed_phases_without_provider_cost"], + "failed_phases_charged_at_bound": state["failed_phases_charged_at_bound"], + }}) + + +def metadata_smoke() -> None: + global openrouter_request_with_metadata_once + body = { + "id": "fake", "model": "fake", "provider": "Google AI Studio", + "choices": [{"finish_reason": "stop", "message": {"content": '{"0": 1, "1": 5}'}}], + "usage": {"cost": 0}, + "openrouter_metadata": { + "requested": "fake", + "endpoints": {"available": [ + {"provider": "Google AI Studio", "model": "fake-release", "selected": True} + ]}, + }, + } + def handler(request: httpx.Request) -> httpx.Response: + assert request.headers["X-OpenRouter-Metadata"] == "enabled" + assert request.headers["Authorization"] == "Bearer test-key" + return httpx.Response(200, json=body) + response = asyncio.run(openrouter_request_with_metadata_once( + {"model": "fake", "messages": [{"role": "user", "content": "test"}]}, + api_key="test-key", transport=httpx.MockTransport(handler))) + assert response["openrouter_metadata"] == body["openrouter_metadata"] + model = MODELS[0] + endpoint = standard_endpoint(model) + exact_release_slug = endpoint["name"].removeprefix(f"{endpoint['provider_name']} | ") + route_response = { + "model": model, + "openrouter_metadata": { + "requested": model, + "endpoints": {"available": [{ + "provider": endpoint["provider_name"], "model": exact_release_slug, "selected": True, + }]}, + }, + } + assert validate_route(route_response, model)["advertised_quantization"] == "unknown" + route_body = body | route_response | {"provider": endpoint["provider_name"]} + + record = OUT / "metadata_capture_smoke.jsonl" + record.unlink(missing_ok=True) + request_count = 0 + async def fake_request(payload: dict) -> dict: + nonlocal request_count + request_count += 1 + return body + kwargs = dict( + n_samples=1, records_path=record, reasoning={"effort": "minimal"}, structured_output=True, + provider=PROVIDER, eval_version=EVAL_VERSION, identity_eval_version=EVAL_VERSION, + seed_schedule=[1], rating_rubric="reversed", request_metadata=True, + ) + items = [{"id": "test", "question": "test", "options": ["a", "b"], "n": 2}] + rows = read_items_rated("fake", items, request_fn=fake_request, **kwargs) + async def must_not_request(payload: dict) -> dict: + raise AssertionError("resume repeated a completed request") + read_items_rated("fake", items, request_fn=must_not_request, reuse_valid_records=True, **kwargs) + events = [json.loads(line) for line in record.read_text().splitlines()] + completed = next(event for event in events if event["event"] == "request_completed") + assert completed["response"]["openrouter_metadata"] == body["openrouter_metadata"] + assert rows[0]["rating_samples_raw"] == [[1.0, 5.0]] + assert rows[0]["rating_samples_transformed"] == [[5.0, 1.0]] + assert request_count == 1 + assert sum(event["event"] == "request_reused" for event in events) == 1 + + rescue_record = OUT / "rescue_seed_capture_smoke.jsonl" + rescue_record.unlink(missing_ok=True) + rescue_payloads = [] + async def fake_rescue_request(payload: dict) -> dict: + rescue_payloads.append(payload) + reply = json.loads(json.dumps(body)) + if len(rescue_payloads) == 1: + reply["choices"][0]["finish_reason"] = "length" + reply["choices"][0]["message"]["content"] = '{"0": 1' + return reply + read_items_rated( + "fake", items, n_samples=1, max_tokens=MAX_TOKENS, + records_path=rescue_record, reasoning={"effort": "high"}, structured_output=True, + provider=PROVIDER, eval_version=EVAL_VERSION, identity_eval_version=EVAL_VERSION, + seed_schedule=[23], rating_rubric="normal", request_metadata=True, + request_fn=fake_rescue_request, + ) + assert len(rescue_payloads) == 2 + assert [payload["seed"] for payload in rescue_payloads] == [23, 23] + rescue_events = [json.loads(line) for line in rescue_record.read_text().splitlines()] + started = [event for event in rescue_events if event["event"] == "request_started"] + assert [event["payload"]["seed"] for event in started] == [23, 23] + + state_before = STATE.read_bytes() if STATE.exists() else None + attempts_before = REQUEST_ATTEMPTS.read_bytes() if REQUEST_ATTEMPTS.exists() else None + initial_state = json.loads(state_before) if state_before is not None else { + "conservative_spent_usd": "0", "provider_reported_spent_usd": "0", + "failed_phases_charged_at_bound": 0, + } + real_request = openrouter_request_with_metadata_once + retry_count = 0 + async def flaky_request(payload: dict, timeout: float = 60.0) -> dict: + nonlocal retry_count + retry_count += 1 + if retry_count == 1: + raise httpx.ReadTimeout("temporary", request=httpx.Request("POST", "https://openrouter.ai")) + return route_body + try: + openrouter_request_with_metadata_once = flaky_request + retry_payload = {"model": MODELS[0], "messages": [{"role": "user", "content": "test"}], + "max_tokens": MAX_TOKENS} + asyncio.run(budgeted_request(retry_payload)) + assert retry_count == 2 + retry_state = json.loads(STATE.read_text()) + spent_delta = (Decimal(retry_state["conservative_spent_usd"]) + - Decimal(initial_state["conservative_spent_usd"])) + assert spent_delta == request_bound(retry_payload) + assert Decimal(retry_state["provider_reported_spent_usd"]) == Decimal( + initial_state["provider_reported_spent_usd"]) + assert retry_state["failed_phases_charged_at_bound"] == ( + initial_state["failed_phases_charged_at_bound"] + 1) + attempt_events = [json.loads(line) for line in REQUEST_ATTEMPTS.read_text().splitlines()] + assert [event["event"] for event in attempt_events[-2:]] == [ + "request_attempt_failed", "request_attempt_completed"] + + invalid_body = json.loads(json.dumps(route_body)) + invalid_body["usage"]["cost"] = 0.001 + invalid_body["openrouter_metadata"]["endpoints"]["available"][0]["provider"] = "Wrong" + async def invalid_route_request(payload: dict, timeout: float = 60.0) -> dict: + return invalid_body + openrouter_request_with_metadata_once = invalid_route_request + before_invalid = Decimal(json.loads(STATE.read_text())["conservative_spent_usd"]) + try: + asyncio.run(budgeted_request(retry_payload)) + except RuntimeError as error: + assert "route mismatch" in str(error) + else: + raise AssertionError("route mismatch did not fail") + invalid_state = json.loads(STATE.read_text()) + assert Decimal(invalid_state["conservative_spent_usd"]) - before_invalid == Decimal("0.001") + invalid_event = json.loads(REQUEST_ATTEMPTS.read_text().splitlines()[-1]) + assert invalid_event["event"] == "request_attempt_route_invalid" + assert invalid_event["response"] == invalid_body + atomic_json(OUT / "route_mismatch_capture_smoke.json", invalid_event) + finally: + openrouter_request_with_metadata_once = real_request + if state_before is None: + STATE.unlink(missing_ok=True) + else: + STATE.write_bytes(state_before) + if attempts_before is None: + REQUEST_ATTEMPTS.unlink(missing_ok=True) + else: + REQUEST_ATTEMPTS.write_bytes(attempts_before) + + atomic_json(OUT / "metadata_capture_smoke.json", { + "status": "passed", "request_header": "X-OpenRouter-Metadata: enabled", + "durable_response_field": completed["response"]["openrouter_metadata"], + "reverse_transform": {"raw": [1.0, 5.0], "transformed": [5.0, 1.0]}, + "resume": "one completed request reused without another call", + "retry": "one retryable timeout preserved, charged conservatively, then completed", + "route_mismatch": "failed inline after charging actual cost and saving the full response", + "route_mismatch_record": str(OUT / "route_mismatch_capture_smoke.json"), + "rescue_seed": "seed 23 present in both initial and rescue payloads", + "rescue_seed_records": str(rescue_record), + "failed_retry_conservative_charge_usd": str(request_bound({"model": MODELS[0], "max_tokens": MAX_TOKENS})), + }) + + +def paid_smoke() -> None: + data = json.loads(MANIFEST.read_text()) + model = data["models"][0] + cell = next(row for row in model["cells"] if row["name"] == "normal_high") + items, _ = rated_items() + smoke_records = REVISED_SMOKE_RECORDS + existing = [json.loads(line) for line in smoke_records.read_text().splitlines()] if smoke_records.exists() else [] + high_runs = {event["run_id"] for event in existing if event.get("event") == "run_started" + and event["settings"]["reasoning"] == {"effort": "high"} + and event["settings"]["max_tokens"] == cell["max_tokens"]} + finished = [event for event in existing if event.get("event") == "run_finished" + and event["run_id"] in high_runs and event["valid_samples"] == 1] + if not finished: + smoke_reservation = reservation_id(SMOKE_RESERVATION) + if not reserve({"id": smoke_reservation, "lane": "google", "reserve_usd": "0.05"}): + raise RuntimeError("global repository cap rejected paid smoke reservation") + before = Decimal(update_state(lambda state: state)["conservative_spent_usd"]) + try: + read_items_rated( + model["id"], [items[0]], n_samples=1, temperature=1.0, max_tokens=cell["max_tokens"], + concurrency=1, req_timeout=REQUEST_TIMEOUT, reasoning=cell["reasoning"], structured_output=True, + records_path=smoke_records, verbose_first=True, provider=PROVIDER, probe_first=True, + eval_version=EVAL_VERSION, identity_eval_version=f"{EVAL_VERSION}-smoke", + seed_schedule=[model["paired_seed_schedule"][0]], rating_rubric="normal", + request_metadata=True, request_fn=budgeted_request, + ) + finally: + after = Decimal(update_state(lambda state: state)["conservative_spent_usd"]) + settle_external_reservation(smoke_reservation, after - before) + events = [json.loads(line) for line in smoke_records.read_text().splitlines()] + high_runs = {event["run_id"] for event in events if event.get("event") == "run_started" + and event["settings"]["reasoning"] == {"effort": "high"} + and event["settings"]["max_tokens"] == cell["max_tokens"]} + finished = [event for event in events if event.get("event") == "run_finished" + and event["run_id"] in high_runs][-1] + run_id = finished["run_id"] + completed = [event for event in events if event.get("event") == "request_completed" + and event["run_id"] == run_id] + parsed = [event for event in events if event.get("event") == "answer_parsed" + and event["run_id"] == run_id][-1] + if finished["valid_samples"] != 1 or finished["failed_samples"] != 0 or parsed["parsed"] is not True: + raise RuntimeError(f"paid smoke was not parse-valid: finished={finished} parsed={parsed}") + phases = [] + for event in completed: + response = event["response"] + message = response["choices"][0]["message"] + phases.append({ + "phase": event["phase"], "route": validate_route(response, model["id"]), + "response_id": response.get("id"), + "finish_reason": response["choices"][0]["finish_reason"], + "content": message.get("content"), "reasoning_text": message.get("reasoning"), + "reasoning_details": message.get("reasoning_details"), + "usage": event.get("usage"), "response": response, + }) + usage_totals = { + "prompt_tokens": sum((event.get("usage") or {})["prompt_tokens"] for event in completed), + "completion_tokens": sum((event.get("usage") or {})["completion_tokens"] for event in completed), + "reasoning_tokens": sum((event.get("usage") or {})["completion_tokens_details"]["reasoning_tokens"] + for event in completed), + "total_tokens": sum((event.get("usage") or {})["total_tokens"] for event in completed), + "provider_reported_cost_usd": str(sum( + (Decimal(str((event.get("usage") or {})["cost"])) for event in completed), Decimal())), + } + state = update_state(lambda value: value) + global_state = json.loads(GLOBAL_BUDGET_STATE.read_text()) + global_external = {key: value for key, value in global_state["external_observed_usd"].items() + if key.startswith(f"{SMOKE_RESERVATION}/")} + global_reservations = {key: value for key, value in global_state["reservations"].items() + if key.startswith(f"{SMOKE_RESERVATION}/")} + atomic_json(REVISED_SMOKE_SUMMARY, { + "status": "passed", "model": model["id"], "cell": cell["name"], + "reasoning": cell["reasoning"], "run_id": run_id, + "protocol_id": finished["protocol_id"], "parse_valid": parsed["parsed"], + "rescued_samples": finished["rescued_samples"], "phases": phases, + "usage_totals": usage_totals, + "budget": { + "provider_reported_spent_usd": state["provider_reported_spent_usd"], + "conservative_spent_usd": state["conservative_spent_usd"], + "reserved_usd": state["reserved_usd"], + "failed_phases_charged_at_bound": state["failed_phases_charged_at_bound"], + }, + "global_accounting": { + "external_observed_usd": global_external, + "matching_reservations": global_reservations, + "total_external_observed_usd": str(sum( + (Decimal(value) for value in global_external.values()), Decimal())), + }, + "records": str(smoke_records), + }) + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--write-manifest", action="store_true") + parser.add_argument("--metadata-smoke", action="store_true") + parser.add_argument("--paid-smoke", action="store_true") + parser.add_argument("--run", action="store_true") + args = parser.parse_args() + if args.write_manifest: + write_manifest() + if args.metadata_smoke: + metadata_smoke() + if args.paid_smoke: + paid_smoke() + if args.run: + run() + + +if __name__ == "__main__": + main() diff --git a/slop/audits/20260918_gemini_flash_high_reasoning_smoke.md b/slop/audits/20260918_gemini_flash_high_reasoning_smoke.md new file mode 100644 index 0000000..4812d9e --- /dev/null +++ b/slop/audits/20260918_gemini_flash_high_reasoning_smoke.md @@ -0,0 +1,128 @@ +# Gemini Flash high-reasoning paid-smoke audit + +Target: the one-item `normal_high` smoke for `google/gemini-3-flash-preview`, run on branch `research/gemini-flash-rubric-v1` from preregistration commit `4955468`. The executed source had uncommitted smoke-selection and route-validation changes that are included in the amended commit reported with this audit. + +| stage | expected | observed | expected? | clues | missing metric | consequence | +|---|---|---|---|---|---|---| +| input and condition | First WVS item, normal rubric, high reasoning | Homosexuality item, `reasoning={"effort":"high"}` | yes | `paid_smoke.jsonl:1-2` | other items | No estimate of item-to-item rescue rate | +| provider routing | Google AI Studio only, exact release visible | `Google AI Studio`, `google/gemini-3-flash-preview-20251217` in both phases | yes | `paid_smoke.jsonl:3,5` | historical quantization | Route is established; quantization remains endpoint-advertised only | +| initial response | Parse-valid rating JSON within 1,024-token budget | `finish_reason="length"`; JSON stopped after key 4 | no | `paid_smoke.jsonl:3-4` | replicate rescue frequency | High reasoning can exhaust this budget | +| rescue | Same conversation returns compact valid JSON | Complete ten-key JSON, `finish_reason="stop"` | yes | `paid_smoke.jsonl:5-6` | none for mechanics | Reader recovered the sample | +| result | One valid sample, failures explicit | valid 1, failed 0, rescued 1 | yes | `paid_smoke.jsonl:7-8` | no coordinate is expected from one item | Mechanics only, not a scientific result | +| usage and reasoning | Full per-phase usage/reasoning retained | 2,362 total tokens, 1,581 reasoning tokens, USD 0.0053585 | yes | `paid_smoke.json:usage_totals,phases` | hidden provider reasoning beyond returned fields | Returned evidence is auditable; hidden fields remain unknowable | +| accounting | Local and global conservative accounting agree; no reservation remains | local actual/conservative USD 0.0053585, global external settlement USD 0.0053585, reserved zero | yes | `paid_smoke.json:budget`; `slop/research/wvs/20260917_score_all_options/budget.json` | none | Spend is reconciled | +| persistence | Raw requests, responses, route, reasoning, usage, parse, and summary saved | JSONL, summary, attempt ledger, and run log exist | yes | experiment directory | exact uncommitted execution diff before amend | Amend commit supplies it | + +## Chronology and primary evidence + +The request used the intended question and rubric. The complete rendered prompt is in `paid_smoke_run.log:4-20`. + +The first response reached the completion limit. `paid_smoke.jsonl:3` records, verbatim inside the response: + +> `"finish_reason": "length"` ... `"content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3,"` ... `"completion_tokens": 1009` ... `"reasoning_tokens": 979` ... `"cost": 0.00311` + +Epistemic context: this is the full provider response saved before parsing, not a later summary. + +The same record reports the selected route: + +> `"requested": "google/gemini-3-flash-preview"` ... `"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true` + +The rescue completed the object. `paid_smoke.jsonl:5` records: + +> `"finish_reason": "stop"` ... `"content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}"` ... `"completion_tokens": 662` ... `"reasoning_tokens": 602` ... `"cost": 0.0022485` + +`paid_smoke.jsonl:6-8` then records `parsed: true`, one valid sample, zero failures, and one rescue. The saved endpoint snapshot contains a present `quantization` field with value `unknown`; this is current endpoint metadata, not historical proof of the serving quantization. + +The run log's only `SHOULD` line says replies should be bare rating JSON with a valid rate near one. `paid_smoke_run.log:21-25` shows the rescued complete JSON and `valid=1/1`; the initial truncation remains visible in the JSONL rather than being summarized away. + +## ML-debug form + +| row | answer | +|---|---| +| log length and config | 26 log lines; full config is `paid_smoke.jsonl:1`, including high reasoning, structured output, provider lock, seed, and timeouts | +| each `SHOULD` and observation | `paid_smoke_run.log:25` expects bare JSON and near-one validity; lines 21-24 show complete rescued JSON and `valid=1/1` | +| null or chance for cited numbers | Not applicable to a transport/parser smoke; no model-quality metric is interpreted | +| initialization, training, gradients, schedule, baseline, held-out | Not applicable; this is one API evaluation request with no training | +| one full sample | Input: `paid_smoke_run.log:4-20`; initial and rescue outputs: `paid_smoke.jsonl:3,5` | +| surprising line | Initial `finish_reason="length"` after 979 reasoning tokens, explained by high reasoning consuming almost all of the initial completion budget | +| missing evidence needed for trust | Rescue frequency across items/releases; one item cannot estimate it | +| diagnoses | See hypotheses below | +| fresh review | Kimi K3's blind read: "Parse -- established only via rescue" and "No" to queueing now; it also found that the panel path does not fail inline on a route mismatch and rescue payloads omit the paired seed | +| cheapest discriminating test | The already-planned full run, stopped by the USD 10 accounting cap, directly measures rescue frequency by condition; queue only after parent smoke review | +| wall clock and GPU | About twelve seconds from reservation to finish in saved timestamps; no GPU | + +The number behind the main diagnosis is 979 reasoning tokens in the initial response. A second cause could be unusually long reasoning for this particular values item, rather than a general property of high reasoning. Rescue counts by item and release separate these explanations. + +## Ranked hypotheses + +### H1 [measurement | Likely | 65%] + +- **Mechanism:** High reasoning often consumes the initial completion budget, so high-reasoning cells invoke the two-phase rescue more often than minimum-reasoning cells. +- **Evidence:** `paid_smoke.jsonl:3` says `finish_reason="length"`, `completion_tokens: 1009`, and `reasoning_tokens: 979`. +- **Contrary evidence:** Only one item and one release were sampled; the rescue stopped normally. +- **Discriminating test:** Compare rescue fractions by cell in the bounded pilot. A high-versus-minimum gap supports the mechanism; similar low rates do not. +- **Fix/action:** Do not change the preregistered protocol before review. Preserve phase counts and let the hard cap stop early if rescues accumulate. +- **Interpretability:** partial; smoke mechanics are interpretable, population rescue frequency is not. + +### H2 [harness | Highly Unlikely | 10%] + +- **Mechanism:** The route checker accepts an alias or wrong provider while presenting it as the intended release. +- **Evidence:** Both phases contain one selected endpoint with provider `Google AI Studio` and exact dated slug `google/gemini-3-flash-preview-20251217`; validation compares all requested, response, provider, and selected-slug fields. +- **Contrary evidence:** The endpoint snapshot is current and supplied the expected dated slug; it does not prove historical quantization. +- **Discriminating test:** Require the same four-field equality on every pilot response, which the runner does. +- **Fix/action:** Keep fail-fast route validation for summary/audit; all raw metadata is already retained. +- **Interpretability:** yes for provider/release routing, no for exact quantization. + +### H3 [bug | Highly Unlikely | 8%] + +- **Mechanism:** A rescue or retry is billed but omitted from local or repository-wide accounting. +- **Evidence:** The raw phases cost USD 0.0031100 and USD 0.0022485; `paid_smoke.json:budget` totals USD 0.0053585 with zero held reservation, and the global external entry records the same amount. +- **Contrary evidence:** This tested successful initial and rescue phases, but not a charged failed attempt in a real request; the offline smoke covers that path. +- **Discriminating test:** Reconcile every future raw `usage.cost` against both ledgers after the run. +- **Fix/action:** Keep per-attempt append-only records and unique global reservation IDs. +- **Interpretability:** yes for this smoke. + +### H4 [data | Likely | 60%] + +- **Mechanism:** This item's reasoning length is not representative of the other items or releases. +- **Evidence:** The smoke contains only Homosexuality, as shown in `paid_smoke_run.log:4-20`. +- **Contrary evidence:** The initial response spent nearly the full fixed completion budget, so the failure mode itself is real even if its frequency is unknown. +- **Discriminating test:** Rescue fractions by item, release, and reasoning cell in the bounded run. +- **Fix/action:** Do not infer a global rescue rate from the smoke. +- **Interpretability:** yes for mechanics, no for frequency. + +## Blind review + +Kimi K3 independently read the runner and complete smoke artifacts without this diagnosis. Its central observation was: + +> **Parse -- established only via rescue.** The initial phase truncated: `"finish_reason": "length"`, `"native_finish_reason": "MAX_TOKENS"` [...] The rescue phase returned `"finish_reason": "stop"` and the full object, yielding `"parse_valid": true`, `"rescued_samples": 1`. + +It recommended not queueing the full run yet. Its additional code findings were that the full panel saves route metadata but does not call `validate_route()` inline, successful provider-reported cost is called conservative rather than tracked under a more exact label, and the rescue follow-up omits the paired seed. Source: reviewer session `/home/code/.pi/agent/sessions/--workspace-2026-lite-moralmaps--/2026-09-18T09-52-42-678Z_01a0b3ee-bcb6-74a9-9466-71816e47002e/cb91d1f9-d68f-4dbe-a960-ffe4aad449a5/run-0/session.jsonl`. + +## Decision + +**Resolve-condition verdict: met.** The required smoke condition was a parse-valid `normal_high` request with exact Google AI Studio routing, exact release metadata, saved reasoning/usage, endpoint quantization field, and reconciled accounting. The raw and summary artifacts contain each field. The smoke also exposed an initial truncation and successful rescue. + +**Prediction check:** The design's scientific predictions about coordinate shifts, rubric sensitivity, family trend, and rubric invariance remain unresolved because the smoke has one item and one cell. The operational prediction that high reasoning might consume the initial budget is supported in this sample. + +**Earliest unsupported link:** One high-reasoning truncation does not establish the rescue frequency or final pilot cost. Condition-level phase counts would support that link. + +**Validity:** Invalid here means that the smoke did not exercise or faithfully record the reviewed route/parser/accounting path. I estimate `P(smoke mechanics are invalid) = 0.03-0.08`; this is a credible positive mechanics smoke, not a scientific result. + +**Three highest-information clues:** + +1. The selected dated release and provider are present in both raw phases, so alias/provider ambiguity did not occur. +2. The first response stopped for length after almost all completion tokens were reasoning, demonstrating the principal high-reasoning risk. +3. Raw phase costs reconcile exactly with local and global accounting, including rescue cost. + +**Missing metrics:** Rescue frequency by condition is highest value, followed by per-item reasoning-token distributions, then whether other releases return the same exact-route metadata shape. + +**Bugs requiring code changes:** Before the full run, the parent should decide whether to add fail-fast route validation to every panel response and copy the paired seed into rescue payloads. The smoke route validation and aggregate phase summaries are already included. `conservative_spent_usd` is exact provider-reported spend for successful phases and a bound only for failed phases, so the field name should not be interpreted as an upper bound on all future billing. + +**Misconceptions requiring reinterpretation:** The USD 9.363456 panel figure is an initial-request bound, not a completion guarantee when rescues occur. The runtime USD 10 stop remains the true spending bound. A rescued high-reasoning cell also tests the repository's two-phase bounded-thinking protocol, not only a single seeded completion. + +**What would change the verdict:** A mismatch between raw phase-cost sums and either ledger, a selected endpoint other than the pinned dated Google release, or a replay that makes another paid call would invalidate the mechanics claim. + +**Recommended sequence:** Amend and push the research branch with the raw smoke and this audit. Wait for parent inspection. Before queueing, decide whether to keep the two-phase high-reasoning protocol and likely early USD 10 stop, or revise and re-preregister its token budget; also resolve panel route validation and rescue seeding. Do not combine these decisions silently. + +-- PI[gpt-5.6-terra] diff --git a/slop/audits/20260918_gemini_flash_high_reasoning_smoke_2048.md b/slop/audits/20260918_gemini_flash_high_reasoning_smoke_2048.md new file mode 100644 index 0000000..fe8c4e0 --- /dev/null +++ b/slop/audits/20260918_gemini_flash_high_reasoning_smoke_2048.md @@ -0,0 +1,92 @@ +# Revised Gemini Flash high-reasoning smoke audit + +Target: the distinct `normal_high`, `max_tokens=2048` smoke for `google/gemini-3-flash-preview` on `research/gemini-flash-rubric-v1`. + +| stage | expected | observed | expected? | source | consequence | +|---|---|---|---|---|---| +| protocol | High reasoning with fixed 2,048 output-token cap | High reasoning, 2,048 in request and protocol identity | yes | `paid_smoke_2048.jsonl:1-2` | Tests the amended condition | +| route | Pinned standard Google AI Studio release | `Google AI Studio`, exact slug `google/gemini-3-flash-preview-20251217` | yes | `paid_smoke_2048.jsonl:3` | Route accepted inline | +| response | Complete ratings without rescue | `finish_reason=stop`, complete ten-key JSON | yes | `paid_smoke_2048.jsonl:3-4` | Requirement met in this sample | +| parser | One valid sample | valid 1, failed 0, rescued 0 | yes | `paid_smoke_2048.jsonl:4-6` | End-to-end reader passed | +| reasoning and usage | Returned reasoning and full usage retained | 166 prompt, 1,535 completion, 1,494 reasoning, 1,701 total tokens | yes | `paid_smoke_2048.json:phases,usage_totals` | Provider-visible evidence retained | +| accounting | Actual cost settled locally and globally; no held reservation | revised smoke USD 0.004688, no matching reservation | yes | `paid_smoke_2048.json:budget,global_accounting` | No observed accounting leak | +| full panel | Must remain unqueued | Only three smoke phases in shared attempt ledger | yes | `request_attempts.jsonl` | Parent must approve before dispatch | + +## Primary evidence + +`paid_smoke_2048.jsonl:3` contains the complete provider response: + +> `"finish_reason": "stop"` ... `"content": "{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}"` ... `"completion_tokens": 1535` ... `"reasoning_tokens": 1494` ... `"cost": 0.004688` + +Epistemic context: this is the full response saved before parsing, not a later reconstruction. + +The same record contains the selected route: + +> `"requested": "google/gemini-3-flash-preview"` ... `"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true` + +`paid_smoke_2048.jsonl:4-6` records `parsed: true`, one valid sample, zero failures, and zero rescues. `paid_smoke_2048_run.log:21-25` quotes the complete raw rating object and `valid=1/1`, satisfying its only `SHOULD` line. + +## ML-debug form + +| row | answer | +|---|---| +| log and config | Complete 25-line log; resolved request settings in `paid_smoke_2048.jsonl:1-2` | +| null or baseline | Not applicable to this transport/parser smoke; no model-quality claim is made | +| one full sample | Prompt in `paid_smoke_2048_run.log:4-20`; raw output in lines 21-22 and JSONL line 3 | +| surprise | Reasoning still used 1,494 tokens, but the response stopped normally and included complete JSON | +| missing evidence | Truncation/rescue frequency across all items and releases | +| wall clock and GPU | About ten seconds from saved reservation/request timestamps; no GPU | +| fresh review | Pi quick oracle found no blocker before parent review and confirmed all 20 cells use 2,048 tokens, route/seed tests pass, accounting reconciles, and the full panel is unqueued | + +The main operational number is zero rescues in one revised sample. A second cause besides the larger budget is sample-to-sample reasoning-length variation. Rescue counts across the full cells separate these explanations. + +## Ranked hypotheses + +### H1 [measurement | Likely | 65%] + +- **Mechanism:** The fixed 2,048-token cap materially reduces condition-specific rescue in high-reasoning cells. +- **Evidence:** `paid_smoke_2048.jsonl:3` stops normally after 1,535 completion tokens, whereas the earlier 1,024-token smoke stopped for length. +- **Contrary evidence:** This is one item and one release. +- **Discriminating test:** Record rescue fractions by cell during the bounded pilot. +- **Action:** Keep phase counts and do not infer a population rate from this smoke. +- **Interpretability:** partial. + +### H2 [harness | Highly Unlikely | 8%] + +- **Mechanism:** A wrong provider/release is accepted despite the requested lock. +- **Evidence:** The route validator matched requested model, response model, selected provider, and dated slug before the caller accepted the response. +- **Contrary evidence:** Only one release has paid confirmation; the other four rely on fail-fast runtime checks. +- **Discriminating test:** The first response for each release must pass the same validator. +- **Action:** Keep inline validation on every paid phase. +- **Interpretability:** yes for this smoke. + +### H3 [bug | Highly Unlikely | 8%] + +- **Mechanism:** A paid phase is omitted from the local or repository-wide ledger. +- **Evidence:** The attempt record reserves USD 0.007168 and settles USD 0.004688; summary and global external accounting record the same actual amount with no held reservation. +- **Contrary evidence:** A real route mismatch was not purchased; that settlement path is covered offline. +- **Discriminating test:** Reconcile all raw `usage.cost` values after any future run. +- **Action:** Retain append-only attempt and global reservation records. +- **Interpretability:** yes for this smoke. + +## Decision + +**Resolve-condition verdict: met.** The revised high-reasoning smoke returned parse-valid JSON without rescue, on the exact pinned route, with complete reasoning/usage and reconciled accounting. + +**Prediction check:** The operational prediction that the larger fixed cap can avoid rescue is supported in this sample. Scientific coordinate, rubric, and release-trend predictions remain unresolved. + +**Earliest unsupported link:** One smoke does not establish the all-item rescue rate; condition-level phase counts are required. + +**Validity:** Invalid means the revised smoke did not exercise or faithfully record the reviewed protocol. I estimate `P(smoke mechanics are invalid) = 0.02-0.06`; this is a credible mechanics result, not a values result. + +**Highest-information clues:** normal stop with complete JSON; exact dated Google route; exact local/global cost reconciliation. + +**Missing metrics:** rescue rate by condition, reasoning-token distribution by item, and first-call confirmation for the other releases. + +**Bugs requiring code changes:** None demonstrated. Both the actual rescue request and its `request_started.payload` contain the paired seed; the offline capture records both phases. + +**Misconceptions requiring reinterpretation:** USD 16.220160 is the initial-request bound, not a rescue/retry worst case. The USD 20 runtime stop is the spending bound. + +**Recommended sequence:** Amend and push the research branch, report this smoke, and keep the full run unqueued until the parent inspects it. + +-- PI[gpt-5.6-terra] diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/budget.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/budget.json new file mode 100644 index 0000000..eba741e --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/budget.json @@ -0,0 +1,10 @@ +{ + "completed_phases": 3, + "completed_phases_without_provider_cost": 0, + "conservative_spent_usd": "0.0100465", + "failed_phases_charged_at_bound": 0, + "hard_cap_usd": "20", + "provider_reported_spent_usd": "0.0100465", + "reserved_usd": "0E-7", + "schema": 1 +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json new file mode 100644 index 0000000..b273839 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json @@ -0,0 +1,31114 @@ +{ + "errors": [], + "fetched_utc": "2026-09-18T10:50:27.498058+00:00", + "models": [ + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Claude" + }, + "created": 1788285838, + "description": "Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors, front-end and visual...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "anthropic/claude-fable-5.1", + "model_name": "Anthropic: Claude Fable 5.1", + "name": "Azure | anthropic/claude-fable-5.1-20260831", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.00000025", + "input_cache_write": "0.0000125", + "input_cache_write_1h": "0.00002", + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "max_completion_tokens", + "tools", + "response_format", + "verbosity", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.88590552977865, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "anthropic/claude-fable-5.1", + "model_name": "Anthropic: Claude Fable 5.1", + "name": "Anthropic | anthropic/claude-fable-5.1-20260831", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.00000025", + "input_cache_write": "0.0000125", + "input_cache_write_1h": "0.00002", + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "Anthropic", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "stop", + "reasoning", + "include_reasoning", + "tools", + "structured_outputs", + "response_format", + "verbosity", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "anthropic", + "throughput_last_30m": null, + "uptime_last_1d": 99.796327090989, + "uptime_last_30m": 99.90160938927542, + "uptime_last_5m": 99.86486486486487 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "anthropic/claude-fable-5.1", + "model_name": "Anthropic: Claude Fable 5.1", + "name": "Amazon Bedrock | anthropic/claude-fable-5.1-20260831", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.00000025", + "input_cache_write": "0.0000125", + "input_cache_write_1h": "0.00002", + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "stop", + "tools", + "response_format", + "verbosity", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "anthropic/claude-fable-5.1", + "model_name": "Anthropic: Claude Fable 5.1", + "name": "Google | anthropic/claude-fable-5.1-20260831", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.00000025", + "input_cache_write": "0.0000125", + "input_cache_write_1h": "0.00002", + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "stop", + "reasoning", + "include_reasoning", + "tools", + "response_format", + "verbosity", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.53641701864633, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "anthropic/claude-fable-5.1", + "name": "Anthropic: Claude Fable 5.1" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1742824755, + "description": "DeepSeek V3, a 685B-parameter, mixture-of-experts model, is the latest iteration of the flagship chat model family from the DeepSeek team. It succeeds the [DeepSeek V3](/deepseek/deepseek-chat-v3) model and performs really well...", + "endpoints": [ + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3-0324", + "model_name": "DeepSeek: DeepSeek V3 0324", + "name": "DeepInfra | deepseek/deepseek-chat-v3-0324", + "pricing": { + "completion": "0.0000009", + "discount": 0, + "input_cache_read": "0.000000135", + "prompt": "0.00000024" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": -5, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "tools", + "tool_choice", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 91.4738801646828, + "uptime_last_30m": 79.4168691426588, + "uptime_last_5m": 80.14981273408239 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3-0324", + "model_name": "DeepSeek: DeepSeek V3 0324", + "name": "SiliconFlow | deepseek/deepseek-chat-v3-0324", + "pricing": { + "completion": "0.000001", + "discount": 0, + "prompt": "0.00000025" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.36748735990433, + "uptime_last_30m": 98.66772402854878, + "uptime_last_5m": 98.96751989675198 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3-0324", + "model_name": "DeepSeek: DeepSeek V3 0324", + "name": "GMICloud | deepseek/deepseek-chat-v3-0324", + "pricing": { + "completion": "0.00000114", + "discount": 0, + "input_cache_read": "0.00000011", + "prompt": "0.00000029" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.78748095270265, + "uptime_last_30m": 95.83777230814444, + "uptime_last_5m": 94.6086956521739 + } + ], + "id": "deepseek/deepseek-chat-v3-0324", + "name": "DeepSeek: DeepSeek V3 0324" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "deepseek-v3.1", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1755779628, + "description": "DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...", + "endpoints": [ + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "DeepInfra | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.00000095", + "discount": 0, + "input_cache_read": "0.00000013", + "prompt": "0.00000025" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.95535433206074, + "uptime_last_30m": 99.96118514684953, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "SiliconFlow | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.000001", + "discount": 0, + "prompt": "0.00000027" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": false, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 95.98386503595546, + "uptime_last_30m": 98.45222072678331, + "uptime_last_5m": 96.6542750929368 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "Novita | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.000000135", + "prompt": "0.00000027" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.9954996810068, + "uptime_last_30m": 99.97104806022004, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "AtlasCloud | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.00000095", + "discount": 0, + "input_cache_read": "0.00000013", + "prompt": "0.0000003" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.25756052609533, + "uptime_last_30m": 99.53408110440034, + "uptime_last_5m": 99.22907488986785 + }, + { + "context_length": 161000, + "latency_last_30m": null, + "max_completion_tokens": 144900, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "CoreWeave | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.00000165", + "discount": 0, + "input_cache_read": "0.00000055", + "prompt": "0.00000055" + }, + "provider_name": "CoreWeave", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.98210303147113, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "Mara | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.0000017", + "discount": 0, + "prompt": "0.0000006" + }, + "provider_name": "Mara", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "top_k", + "top_p", + "response_format", + "logprobs", + "top_logprobs", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "mara", + "throughput_last_30m": null, + "uptime_last_1d": 95.68618353901827, + "uptime_last_30m": 96.9377675337504, + "uptime_last_5m": 95.44364508393285 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "Google | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.0000017", + "discount": 0, + "prompt": "0.0000006" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us-west2", + "throughput_last_30m": null, + "uptime_last_1d": 0, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 7168, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-chat-v3.1", + "model_name": "DeepSeek: DeepSeek V3.1", + "name": "SambaNova | deepseek/deepseek-chat-v3.1", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "prompt": "0.00000065" + }, + "provider_name": "SambaNova", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sambanova/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.4269592037011, + "uptime_last_30m": 98.59719438877755, + "uptime_last_5m": 99.19354838709677 + } + ], + "id": "deepseek/deepseek-chat-v3.1", + "name": "DeepSeek: DeepSeek V3.1" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1764594642, + "description": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...", + "endpoints": [ + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "GMICloud | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000003096", + "discount": 0.28, + "input_cache_read": "0.0000000216", + "prompt": "0.0000002088" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.42327898564952, + "uptime_last_30m": 99.9373727884766, + "uptime_last_5m": 99.91978609625669 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 64000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "StreamLake | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000032175", + "discount": 0.25, + "input_cache_read": "0.00000002145", + "prompt": "0.0000002145" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "max_tokens", + "temperature", + "top_p", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.08766090868617, + "uptime_last_30m": 99.62080173347779, + "uptime_last_5m": 99.70344009489916 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "SiliconFlow | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000042", + "discount": 0, + "input_cache_read": "0.000000135", + "prompt": "0.000000259" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.7704027418172, + "uptime_last_30m": 99.00045433893685, + "uptime_last_5m": 99.41812136325852 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "DeepInfra | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000038", + "discount": 0, + "input_cache_read": "0.00000013", + "prompt": "0.00000026" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tool_choice", + "tools", + "response_format", + "logit_bias", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.34048714293215, + "uptime_last_30m": 98.87940234791888, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "AtlasCloud | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000038", + "discount": 0, + "input_cache_read": "0.00000013", + "prompt": "0.00000026" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "response_format", + "tools", + "tool_choice", + "seed", + "logit_bias", + "stop", + "min_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.5665922967796, + "uptime_last_30m": 99.74011348377874, + "uptime_last_5m": 99.73652431660994 + }, + { + "context_length": 160000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Venice | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000039024", + "discount": 0.187, + "input_cache_read": "0.00000013008", + "prompt": "0.00000026829" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "structured_outputs", + "response_format", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 99.65051436835128, + "uptime_last_30m": 99.42738157209786, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Novita | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.0000001345", + "prompt": "0.000000269" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": false, + "function": false, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.98664480954452, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Baidu | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000042", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000028" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "stop", + "seed", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.4259386909648, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "DigitalOcean | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000096", + "discount": 0, + "input_cache_read": "0.00000009", + "prompt": "0.0000003" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 98.6027391766428, + "uptime_last_30m": 95.23589269195189, + "uptime_last_5m": 99.95440036479708 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 98304, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Alibaba | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000011115", + "discount": 0, + "input_cache_read": "0.0000000741", + "input_cache_write": "0.00000046345", + "prompt": "0.0000003705" + }, + "provider_name": "Alibaba", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "structured_outputs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.35222245628819, + "uptime_last_30m": 98.96519285042334, + "uptime_last_5m": 99.13793103448276 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Friendli | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.0000005" + }, + "provider_name": "Friendli", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "friendli", + "throughput_last_30m": null, + "uptime_last_1d": 99.97858601544941, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Google | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.00000168", + "discount": 0, + "prompt": "0.00000056" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex", + "throughput_last_30m": null, + "uptime_last_1d": 98.5853536615846, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 64000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Phala | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.0000005", + "prompt": "0.000001" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "structured_outputs", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.76762297141134, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 32768, + "latency_last_30m": null, + "max_completion_tokens": 7168, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "Mara | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000045", + "discount": 0, + "prompt": "0.000003" + }, + "provider_name": "Mara", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "top_k", + "top_p", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "mara", + "throughput_last_30m": null, + "uptime_last_1d": 85.12820512820512, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 32768, + "latency_last_30m": null, + "max_completion_tokens": 7168, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2", + "model_name": "DeepSeek: DeepSeek V3.2", + "name": "SambaNova | deepseek/deepseek-v3.2-20251201", + "pricing": { + "completion": "0.0000045", + "discount": 0, + "prompt": "0.000003" + }, + "provider_name": "SambaNova", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "sambanova", + "throughput_last_30m": null, + "uptime_last_1d": 98.58078602620087, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "deepseek/deepseek-v3.2", + "name": "DeepSeek: DeepSeek V3.2" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "deepseek-v3.1", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1759150481, + "description": "DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...", + "endpoints": [ + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2-exp", + "model_name": "DeepSeek: DeepSeek V3.2 Exp", + "name": "SiliconFlow | deepseek/deepseek-v3.2-exp", + "pricing": { + "completion": "0.00000041", + "discount": 0, + "prompt": "0.00000027" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.59946279368518, + "uptime_last_30m": 99.48051948051948, + "uptime_last_5m": 99.6742671009772 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 147456, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2-exp", + "model_name": "DeepSeek: DeepSeek V3.2 Exp", + "name": "AtlasCloud | deepseek/deepseek-v3.2-exp", + "pricing": { + "completion": "0.00000041", + "discount": 0, + "input_cache_read": "0.00000027", + "prompt": "0.00000027" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.02096823928461, + "uptime_last_30m": 99.80430528375733, + "uptime_last_5m": 98.46153846153847 + }, + { + "context_length": 163840, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v3.2-exp", + "model_name": "DeepSeek: DeepSeek V3.2 Exp", + "name": "Novita | deepseek/deepseek-v3.2-exp", + "pricing": { + "completion": "0.00000041", + "discount": 0, + "prompt": "0.00000027" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.5415029177087, + "uptime_last_30m": 99.94148624926858, + "uptime_last_5m": 99.45945945945947 + } + ], + "id": "deepseek/deepseek-v3.2-exp", + "name": "DeepSeek: DeepSeek V3.2 Exp" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1777000666, + "description": "DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "OpenInference | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000014", + "discount": 0, + "input_cache_read": "0.000000014", + "prompt": "0.00000005" + }, + "provider_name": "OpenInference", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "stop", + "seed", + "tools", + "tool_choice", + "response_format", + "top_p", + "frequency_penalty", + "presence_penalty", + "top_k", + "min_p", + "logit_bias", + "repetition_penalty", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "open-inference/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.8762290614328, + "uptime_last_30m": 99.99724016117459, + "uptime_last_5m": 99.995037959609 + }, + { + "context_length": 1024000, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "StreamLake | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.000000177212", + "discount": 0.3671, + "input_cache_read": "0.0000000177212", + "prompt": "0.000000088606" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "structured_outputs", + "max_tokens", + "temperature", + "top_p", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.76605386006817, + "uptime_last_30m": 97.9149584163055, + "uptime_last_5m": 98.57287449392712 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "DeepInfra | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000018", + "discount": 0, + "input_cache_read": "0.000000018", + "prompt": "0.00000009" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.64577416959625, + "uptime_last_30m": 99.86004795917496, + "uptime_last_5m": 99.92922403809034 + }, + { + "context_length": 1048575, + "latency_last_30m": null, + "max_completion_tokens": 943717, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "GMICloud | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.000000182", + "discount": 0.35, + "input_cache_read": "0.0000000182", + "prompt": "0.000000091" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.95430398207432, + "uptime_last_30m": 99.99643687478287, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Venice | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.0000001925", + "discount": 0.3, + "input_cache_read": "0.0000000196", + "prompt": "0.0000000966" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 96.95928374961021, + "uptime_last_30m": 99.8238644147406, + "uptime_last_5m": 99.88586646376261 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "DigitalOcean | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.000000196", + "discount": 0, + "input_cache_read": "0.0000000196", + "prompt": "0.000000098" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "tool_choice", + "tools", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 99.93511668778355, + "uptime_last_30m": 99.9732963042085, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "SiliconFlow | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000013" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.8390144261458, + "uptime_last_30m": 97.27985246657445, + "uptime_last_5m": 97.88732394366197 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": 1000000, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Alibaba | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.000000268", + "discount": 0, + "input_cache_read": "0.0000000268", + "prompt": "0.000000134" + }, + "provider_name": "Alibaba", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "structured_outputs", + "top_k", + "frequency_penalty", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.15166393515693, + "uptime_last_30m": 99.41332092202924, + "uptime_last_5m": 99.2 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Baidu | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000014" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.7402817735906, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Novita | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000014" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tool_choice", + "tools", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.84600233233773, + "uptime_last_30m": 99.9974189885016, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "AtlasCloud | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000014" + }, + "provider_name": "AtlasCloud", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.97115257969953, + "uptime_last_30m": 99.9880820432145, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Parasail | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.00000007", + "prompt": "0.00000014" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.66546726636682, + "uptime_last_30m": 99.89439853076216, + "uptime_last_5m": 99.7881010594947 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "NextBit | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000035", + "discount": 0, + "input_cache_read": "0.000000035", + "prompt": "0.00000015" + }, + "provider_name": "NextBit", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "repetition_penalty", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nextbit/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.95291081842996, + "uptime_last_30m": 99.89203023153517, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Mancer 2 | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "prompt": "0.00000019" + }, + "provider_name": "Mancer 2", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "logit_bias", + "top_k", + "min_p", + "seed", + "top_a", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "structured_outputs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "mancer/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.6342645991607, + "uptime_last_30m": 98.9267461669506, + "uptime_last_5m": 99.3103448275862 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Phala | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.00000007", + "prompt": "0.0000002" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.59386758201862, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash", + "model_name": "DeepSeek: DeepSeek V4 Flash 0423", + "name": "Azure | deepseek/deepseek-v4-flash-20260423", + "pricing": { + "completion": "0.00000056", + "discount": 0, + "input_cache_read": "0.000000031", + "prompt": "0.00000021" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "max_completion_tokens", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": 87.62834961182068, + "uptime_last_30m": 79.30675909878683, + "uptime_last_5m": null + } + ], + "id": "deepseek/deepseek-v4-flash", + "name": "DeepSeek: DeepSeek V4 Flash 0423" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1785478908, + "description": "DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Inceptron | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.0000001767", + "discount": 0, + "input_cache_read": "0.0000000088", + "prompt": "0.0000000558" + }, + "provider_name": "Inceptron", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "top_p", + "stop", + "frequency_penalty", + "logit_bias", + "parallel_tool_calls", + "presence_penalty", + "seed", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inceptron/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 90.10607220488347, + "uptime_last_30m": 92.79766123316796, + "uptime_last_5m": 69.71428571428572 + }, + { + "context_length": 1024000, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "StreamLake | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000178728", + "discount": 0.8646, + "input_cache_read": "0.0000000018956", + "prompt": "0.000000059576" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.52536464787705, + "uptime_last_30m": 96.93141529809691, + "uptime_last_5m": 96.97695035460993 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Relace | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000012", + "discount": 0, + "input_cache_read": "0.000000012", + "prompt": "0.00000006" + }, + "provider_name": "Relace", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "tools", + "tool_choice", + "temperature", + "top_p", + "top_k", + "min_p", + "stop", + "max_tokens", + "logit_bias", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "relace/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.67517071964039, + "uptime_last_30m": 99.60973859214319, + "uptime_last_5m": 99.66627479236982 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "DeepInfra | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000018", + "discount": 0, + "input_cache_read": "0.000000015", + "prompt": "0.00000006" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.79440604062533, + "uptime_last_30m": 99.66781769926656, + "uptime_last_5m": 99.7208729065468 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Sail Research | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000342", + "discount": 0, + "input_cache_read": "0.0000000228", + "prompt": "0.0000000741" + }, + "provider_name": "Sail Research", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "frequency_penalty", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_k", + "top_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/us", + "throughput_last_30m": null, + "uptime_last_1d": 96.5024761261463, + "uptime_last_30m": 94.156383483626, + "uptime_last_5m": 95.93331649406416 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Sail Research | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000342", + "discount": 0, + "input_cache_read": "0.0000000228", + "prompt": "0.0000000741" + }, + "provider_name": "Sail Research", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "max_tokens", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 96.69247800450357, + "uptime_last_30m": 95.01443859351113, + "uptime_last_5m": 95.1837416481069 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Makora | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000195", + "discount": 0, + "input_cache_read": "0.0000000196", + "prompt": "0.00000009" + }, + "provider_name": "Makora", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "max_tokens", + "seed", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "makora", + "throughput_last_30m": null, + "uptime_last_1d": 99.07804812008007, + "uptime_last_30m": 99.46066414085135, + "uptime_last_5m": 99.4413407821229 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Wafer | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000025", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.0000001" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.93528452895471, + "uptime_last_30m": 99.95210118624406, + "uptime_last_5m": 99.96968012532214 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Reka | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000066", + "discount": 0, + "input_cache_read": "0.000000007", + "prompt": "0.00000011" + }, + "provider_name": "Reka", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "reka/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.84619695808182, + "uptime_last_30m": 99.81021182808864, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "DigitalOcean | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000238", + "discount": 0, + "input_cache_read": "0.0000000238", + "prompt": "0.000000119" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 99.9158980462164, + "uptime_last_30m": 99.97578497053838, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "BaseTen | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000026", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000013" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.94414898875071, + "uptime_last_30m": 99.8991935483871, + "uptime_last_5m": 99.85734664764621 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "BaseTen | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000026", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000013" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.93597118703417, + "uptime_last_30m": 99.93666877770741, + "uptime_last_5m": 99.66329966329967 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "CoreWeave | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.00000007", + "prompt": "0.00000013" + }, + "provider_name": "CoreWeave", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.99466860814111, + "uptime_last_30m": 99.99395631625389, + "uptime_last_5m": 99.99277769752997 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Together | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000014" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.68454536255284, + "uptime_last_30m": 99.84616229877479, + "uptime_last_5m": 99.60068454078723 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Parasail | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000014" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.68762446756148, + "uptime_last_30m": 99.9586659995214, + "uptime_last_5m": 99.98627693152189 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Morph | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000399625", + "discount": 0, + "input_cache_read": "0.0000000359375", + "prompt": "0.000000141953" + }, + "provider_name": "Morph", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "morph/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 97.65944360075551, + "uptime_last_30m": 99.82449982449982, + "uptime_last_5m": 99.79674796747967 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Venice | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000035", + "discount": 0, + "input_cache_read": "0.000000035", + "prompt": "0.000000175" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 99.46218210726646, + "uptime_last_30m": 99.50706228078491, + "uptime_last_5m": 99.78880675818374 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Mancer 2 | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Mancer 2", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "logit_bias", + "top_k", + "min_p", + "seed", + "top_a", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "mancer/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.43448965236385, + "uptime_last_30m": 99.02439024390245, + "uptime_last_5m": 99.7566909975669 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Fireworks | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000066", + "discount": 0, + "input_cache_read": "0.000000007", + "prompt": "0.00000022" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 97.3099168075304, + "uptime_last_30m": 98.6602158541124, + "uptime_last_5m": 99.74226804123711 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "SiliconFlow | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000066", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000022" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.62260075198091, + "uptime_last_30m": 99.85525601592184, + "uptime_last_5m": 100 + }, + { + "context_length": 1048575, + "latency_last_30m": null, + "max_completion_tokens": 943717, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "GMICloud | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000000858", + "discount": 0.35, + "input_cache_read": "0.0000000091", + "prompt": "0.000000286" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.87872949206174, + "uptime_last_30m": 99.25084217406607, + "uptime_last_5m": 99.83198924731182 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "NextBit | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000001056", + "discount": 0, + "input_cache_read": "0.000000012", + "prompt": "0.000000352" + }, + "provider_name": "NextBit", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "repetition_penalty", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nextbit/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.14591858351824, + "uptime_last_30m": 99.93626513702995, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Alibaba | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.000001056", + "discount": 0, + "input_cache_read": "0.0000000352", + "overrides": [ + { + "completion": "0.000001056", + "input_cache_read": "0.0000000352", + "prompt": "0.000000352", + "utc_end": 1400, + "utc_start": 0 + }, + { + "completion": "0.000000528", + "input_cache_read": "0.0000000176", + "prompt": "0.000000176", + "utc_end": 0, + "utc_start": 1400 + } + ], + "prompt": "0.000000352" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "top_k", + "frequency_penalty", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.998369612475, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Novita | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.0000012276", + "discount": 0.07, + "input_cache_read": "0.00000002604", + "prompt": "0.0000004092" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "response_format", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.97899241392724, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Phala | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000044" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.80183285626428, + "uptime_last_30m": 99.95596653456627, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Baidu | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000014", + "prompt": "0.00000044" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.97252736913808, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "AtlasCloud | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000028", + "prompt": "0.00000044" + }, + "provider_name": "AtlasCloud", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.85656735120804, + "uptime_last_30m": 99.82608695652175, + "uptime_last_5m": 100 + }, + { + "context_length": 1310720, + "latency_last_30m": null, + "max_completion_tokens": 1179648, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4-flash-0731", + "model_name": "DeepSeek: DeepSeek V4 Flash 0731", + "name": "Cloudflare | deepseek/deepseek-v4-flash-20260731", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000014", + "prompt": "0.00000044" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "min_p", + "stop", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.98624968151836, + "uptime_last_30m": 99.96940024479805, + "uptime_last_5m": 100 + } + ], + "id": "deepseek/deepseek-v4-flash-0731", + "name": "DeepSeek: DeepSeek V4 Flash 0731" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "DeepSeek" + }, + "created": 1789021285, + "description": "DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Relace | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000015", + "prompt": "0.00000015" + }, + "provider_name": "Relace", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "stop", + "max_tokens", + "logit_bias", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "relace/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.01856475661344, + "uptime_last_30m": 98.42200925313945, + "uptime_last_5m": 97.98499464094319 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "DeepSeek | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000003", + "overrides": [ + { + "completion": "0.0000006", + "input_cache_read": "0.000000003", + "prompt": "0.00000015", + "utc_days": [ + "saturday", + "sunday" + ] + }, + { + "completion": "0.0000006", + "input_cache_read": "0.000000003", + "prompt": "0.00000015", + "utc_days": [ + "monday", + "tuesday", + "wednesday", + "thursday", + "friday" + ], + "utc_end": 100, + "utc_start": 0 + }, + { + "completion": "0.0000012", + "input_cache_read": "0.000000006", + "prompt": "0.0000003", + "utc_days": [ + "monday", + "tuesday", + "wednesday", + "thursday", + "friday" + ], + "utc_end": 400, + "utc_start": 100 + }, + { + "completion": "0.0000006", + "input_cache_read": "0.000000003", + "prompt": "0.00000015", + "utc_days": [ + "monday", + "tuesday", + "wednesday", + "thursday", + "friday" + ], + "utc_end": 600, + "utc_start": 400 + }, + { + "completion": "0.0000012", + "input_cache_read": "0.000000006", + "prompt": "0.0000003", + "utc_days": [ + "monday", + "tuesday", + "wednesday", + "thursday", + "friday" + ], + "utc_end": 1000, + "utc_start": 600 + }, + { + "completion": "0.0000006", + "input_cache_read": "0.000000003", + "prompt": "0.00000015", + "utc_days": [ + "monday", + "tuesday", + "wednesday", + "thursday", + "friday" + ], + "utc_end": 0, + "utc_start": 1000 + } + ], + "prompt": "0.00000015" + }, + "provider_name": "DeepSeek", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepseek", + "throughput_last_30m": null, + "uptime_last_1d": 99.97854678998357, + "uptime_last_30m": 99.99596507367775, + "uptime_last_5m": 99.99799498746867 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Morph | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.00000077988", + "discount": 0.3501, + "input_cache_read": "0.0000000058491", + "prompt": "0.00000019497" + }, + "provider_name": "Morph", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "morph/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.69802922400619, + "uptime_last_30m": 99.54379078032996, + "uptime_last_5m": 99.07834101382488 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Wafer | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000002" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer", + "throughput_last_30m": null, + "uptime_last_1d": 99.24837611159492, + "uptime_last_30m": 99.52144303331492, + "uptime_last_5m": 98.98624185372918 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "DeepInfra | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000002" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.74938546817228, + "uptime_last_30m": 99.8757632374752, + "uptime_last_5m": 99.88265888621083 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Fireworks | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.00000066", + "discount": 0, + "input_cache_read": "0.000000007", + "prompt": "0.00000022" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.6817305680782, + "uptime_last_30m": 99.84815252235532, + "uptime_last_5m": 99.72574562907096 + }, + { + "context_length": 1024000, + "latency_last_30m": null, + "max_completion_tokens": 384000, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "StreamLake | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.000001128", + "discount": 0.06, + "input_cache_read": "0.00000000564", + "prompt": "0.000000282" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "logprobs", + "top_logprobs", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.28310774088357, + "uptime_last_30m": 99.2784683526218, + "uptime_last_5m": 99.572877736252 + }, + { + "context_length": 1048575, + "latency_last_30m": null, + "max_completion_tokens": 943717, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "GMICloud | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.00000114", + "discount": 0.05, + "input_cache_read": "0.0000000057", + "prompt": "0.000000285" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.85035960380924, + "uptime_last_30m": 99.98113800691605, + "uptime_last_5m": 99.95911692559281 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Novita | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.00000114", + "discount": 0.05, + "input_cache_read": "0.0000000057", + "prompt": "0.000000285" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.94962259884417, + "uptime_last_30m": 99.99756507341304, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "AtlasCloud | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000003" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.8602853100265, + "uptime_last_30m": 99.93187237424776, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "BaseTen | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000003" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.90541594194207, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Makora | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000003" + }, + "provider_name": "Makora", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "max_tokens", + "seed", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "makora", + "throughput_last_30m": null, + "uptime_last_1d": 97.71141868227971, + "uptime_last_30m": 99.6967859308672, + "uptime_last_5m": 99.29178470254958 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "DigitalOcean | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000003" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "response_format", + "logprobs", + "top_logprobs", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 97.6188391457766, + "uptime_last_30m": 99.37903941198834, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Alibaba | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000003", + "overrides": [ + { + "completion": "0.0000012", + "input_cache_read": "0.00000003", + "prompt": "0.0000003", + "utc_end": 1400, + "utc_start": 0 + }, + { + "completion": "0.0000006", + "input_cache_read": "0.000000015", + "prompt": "0.00000015", + "utc_end": 0, + "utc_start": 1400 + } + ], + "prompt": "0.0000003" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "top_k", + "frequency_penalty", + "stop", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.78518266087181, + "uptime_last_30m": 99.95008111818295, + "uptime_last_5m": 99.93714644877436 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Together | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000003" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.80891825792438, + "uptime_last_30m": 99.93863661720405, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "SiliconFlow | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000003" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96377890010693, + "uptime_last_30m": 99.98777655543331, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Modal | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000003" + }, + "provider_name": "Modal", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "modal", + "throughput_last_30m": null, + "uptime_last_1d": 98.27177761054259, + "uptime_last_30m": 99.5563699488861, + "uptime_last_5m": 99.7780244173141 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "BaseTen | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000003" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.83798817602005, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Parasail | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.000000006", + "prompt": "0.0000003" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.05975476848319, + "uptime_last_30m": 90.75663026521062, + "uptime_last_5m": 88.10289389067523 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 393216, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Phala | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.00000138", + "discount": 0, + "input_cache_read": "0.0000000069", + "prompt": "0.000000345" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.7435376223871, + "uptime_last_30m": 99.82445874780574, + "uptime_last_5m": 99.83349191246432 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "deepseek/deepseek-v4.1-flash", + "model_name": "DeepSeek: DeepSeek V4.1 Flash", + "name": "Venice | deepseek/deepseek-v4.1-flash-20260910", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.0000000075", + "prompt": "0.000000375" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.60742806055187, + "uptime_last_30m": 99.84690753214942, + "uptime_last_5m": 100 + } + ], + "id": "deepseek/deepseek-v4.1-flash", + "name": "DeepSeek: DeepSeek V4.1 Flash" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text", + "audio", + "video" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1750172488, + "description": "Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and scientific tasks. It includes built-in \"thinking\" capabilities, enabling it to provide responses with greater...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google | google/gemini-2.5-flash", + "pricing": { + "audio": "0.000001", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/eu", + "throughput_last_30m": null, + "uptime_last_1d": 99.43693536712865, + "uptime_last_30m": 99.37168904767772, + "uptime_last_5m": 99.57033016734509 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google | google/gemini-2.5-flash", + "pricing": { + "audio": "0.000001", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.36664593625491, + "uptime_last_30m": 99.54002579607697, + "uptime_last_5m": 99.7497303128371 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google | google/gemini-2.5-flash", + "pricing": { + "audio": "0.0000018", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000054", + "input_audio_cache": "0.00000018", + "input_cache_read": "0.000000054", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000045", + "prompt": "0.00000054", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.89669011006477, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google AI Studio | google/gemini-2.5-flash", + "pricing": { + "audio": "0.0000005", + "completion": "0.00000125", + "discount": 0, + "image": "0.00000015", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000015", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000125", + "prompt": "0.00000015", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.99188650377857, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google AI Studio | google/gemini-2.5-flash", + "pricing": { + "audio": "0.000001", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.93823623761332, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google | google/gemini-2.5-flash", + "pricing": { + "audio": "0.000001", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex", + "throughput_last_30m": null, + "uptime_last_1d": 86.20208781576221, + "uptime_last_30m": 79.48523845571536, + "uptime_last_5m": 74.19354838709677 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash", + "model_name": "Google: Gemini 2.5 Flash", + "name": "Google AI Studio | google/gemini-2.5-flash", + "pricing": { + "audio": "0.0000018", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000054", + "input_audio_cache": "0.00000018", + "input_cache_read": "0.000000054", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000045", + "prompt": "0.00000054", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.43181818181817, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "google/gemini-2.5-flash", + "name": "Google: Gemini 2.5 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file", + "audio", + "video" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1753200276, + "description": "Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash-lite", + "model_name": "Google: Gemini 2.5 Flash Lite", + "name": "Google | google/gemini-2.5-flash-lite", + "pricing": { + "audio": "0.0000003", + "completion": "0.0000004", + "discount": 0, + "image": "0.0000001", + "input_audio_cache": "0.00000003", + "input_cache_read": "0.00000001", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000004", + "prompt": "0.0000001", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/eu", + "throughput_last_30m": null, + "uptime_last_1d": 99.85968110567643, + "uptime_last_30m": 99.95496694164126, + "uptime_last_5m": 99.95020121401365 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash-lite", + "model_name": "Google: Gemini 2.5 Flash Lite", + "name": "Google | google/gemini-2.5-flash-lite", + "pricing": { + "audio": "0.0000003", + "completion": "0.0000004", + "discount": 0, + "image": "0.0000001", + "input_audio_cache": "0.00000003", + "input_cache_read": "0.00000001", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000004", + "prompt": "0.0000001", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex", + "throughput_last_30m": null, + "uptime_last_1d": 98.84689329375523, + "uptime_last_30m": 95.20199245504158, + "uptime_last_5m": 99.19427036705461 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash-lite", + "model_name": "Google: Gemini 2.5 Flash Lite", + "name": "Google AI Studio | google/gemini-2.5-flash-lite", + "pricing": { + "audio": "0.00000015", + "completion": "0.0000002", + "discount": 0, + "image": "0.00000005", + "input_audio_cache": "0.000000015", + "input_cache_read": "0.000000005", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.0000002", + "prompt": "0.00000005", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.98170875244149, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash-lite", + "model_name": "Google: Gemini 2.5 Flash Lite", + "name": "Google AI Studio | google/gemini-2.5-flash-lite", + "pricing": { + "audio": "0.0000003", + "completion": "0.0000004", + "discount": 0, + "image": "0.0000001", + "input_audio_cache": "0.00000003", + "input_cache_read": "0.00000001", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000004", + "prompt": "0.0000001", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.29871743360194, + "uptime_last_30m": 99.49196103042257, + "uptime_last_5m": 99.22720247295209 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "google/gemini-2.5-flash-lite", + "model_name": "Google: Gemini 2.5 Flash Lite", + "name": "Google AI Studio | google/gemini-2.5-flash-lite", + "pricing": { + "audio": "0.00000054", + "completion": "0.00000072", + "discount": 0, + "image": "0.00000018", + "input_audio_cache": "0.000000054", + "input_cache_read": "0.000000018", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.00000072", + "prompt": "0.00000018", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "google/gemini-2.5-flash-lite", + "name": "Google: Gemini 2.5 Flash Lite" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file", + "audio", + "video" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1765987078, + "description": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.0000005", + "completion": "0.0000015", + "discount": 0, + "image": "0.00000025", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000025", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.0000015", + "prompt": "0.00000025", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tool_choice", + "tools", + "stop", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 23.440166887919588, + "uptime_last_30m": 5.683269476372925, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google AI Studio | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.0000005", + "completion": "0.0000015", + "discount": 0, + "image": "0.00000025", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000025", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.0000015", + "prompt": "0.00000025", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.87996625138082, + "uptime_last_30m": 99.92729035385362, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google AI Studio | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.000001", + "completion": "0.000003", + "discount": 0, + "image": "0.0000005", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000005", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.000003", + "prompt": "0.0000005", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.93248603462276, + "uptime_last_30m": 99.97724324693353, + "uptime_last_5m": 99.98356344510191 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.000001", + "completion": "0.000003", + "discount": 0, + "image": "0.0000005", + "input_audio_cache": "0.0000001", + "input_cache_read": "0.00000005", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.000003", + "prompt": "0.0000005", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tool_choice", + "tools", + "stop", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 97.98778860358212, + "uptime_last_30m": 98.18275912259233, + "uptime_last_5m": 99.06388452193798 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.0000018", + "completion": "0.0000054", + "discount": 0, + "image": "0.0000009", + "input_audio_cache": "0.00000018", + "input_cache_read": "0.00000009", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000054", + "prompt": "0.0000009", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tool_choice", + "tools", + "stop", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.7975951777051, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3-flash-preview", + "model_name": "Google: Gemini 3 Flash Preview", + "name": "Google AI Studio | google/gemini-3-flash-preview-20251217", + "pricing": { + "audio": "0.0000018", + "completion": "0.0000054", + "discount": 0, + "image": "0.0000009", + "input_audio_cache": "0.00000018", + "input_cache_read": "0.00000009", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000054", + "prompt": "0.0000009", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.94732221246707, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3-flash-preview", + "name": "Google: Gemini 3 Flash Preview" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1778168828, + "description": "Gemini 3.1 Flash Lite is Google\u2019s GA high-efficiency multimodal model optimized for low-latency, high-volume workloads. It supports text, image, video, audio, and PDF inputs, and is designed for lightweight agentic...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.00000025", + "completion": "0.00000075", + "discount": 0, + "image": "0.000000125", + "input_audio_cache": "0.000000025", + "input_cache_read": "0.0000000125", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000075", + "prompt": "0.000000125", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 25.3209242618742, + "uptime_last_30m": 41.32231404958678, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.00000025", + "completion": "0.00000075", + "discount": 0, + "image": "0.000000125", + "input_audio_cache": "0.000000025", + "input_cache_read": "0.0000000125", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000075", + "prompt": "0.000000125", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.97566440947229, + "uptime_last_30m": 99.97165934533088, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.0000005", + "completion": "0.0000015", + "discount": 0, + "image": "0.00000025", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000025", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000015", + "prompt": "0.00000025", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.8966060706413, + "uptime_last_30m": 99.8388434648655, + "uptime_last_5m": 99.84995225753649 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.0000005", + "completion": "0.0000015", + "discount": 0, + "image": "0.00000025", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000025", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000015", + "prompt": "0.00000025", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.82247206217116, + "uptime_last_30m": 99.95438489789234, + "uptime_last_5m": 99.96823147039417 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.00000055", + "completion": "0.00000165", + "discount": 0, + "image": "0.000000275", + "input_audio_cache": "0.000000055", + "input_cache_read": "0.0000000275", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.00000165", + "prompt": "0.000000275", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/eu", + "throughput_last_30m": null, + "uptime_last_1d": 99.92898848035348, + "uptime_last_30m": 99.62894248608535, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.00000055", + "completion": "0.00000165", + "discount": 0, + "image": "0.000000275", + "input_audio_cache": "0.000000055", + "input_cache_read": "0.0000000275", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.00000165", + "prompt": "0.000000275", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us", + "throughput_last_30m": null, + "uptime_last_1d": 99.89496959646213, + "uptime_last_30m": 98.69281045751634, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.0000009", + "completion": "0.0000027", + "discount": 0, + "image": "0.00000045", + "input_audio_cache": "0.00000009", + "input_cache_read": "0.000000045", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000027", + "prompt": "0.00000045", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.48685726254058, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite", + "model_name": "Google: Gemini 3.1 Flash Lite", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-20260507", + "pricing": { + "audio": "0.0000009", + "completion": "0.0000027", + "discount": 0, + "image": "0.00000045", + "input_audio_cache": "0.00000009", + "input_cache_read": "0.000000045", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000027", + "prompt": "0.00000045", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.89876903142209, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "google/gemini-3.1-flash-lite", + "name": "Google: Gemini 3.1 Flash Lite" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1772512673, + "description": "Gemini 3.1 Flash Lite Preview is Google's high-efficiency model optimized for high-volume use cases. It outperforms Gemini 2.5 Flash Lite on overall quality and approaches Gemini 2.5 Flash performance across...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite-preview", + "model_name": "Google: Gemini 3.1 Flash Lite Preview", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-preview-20260303", + "pricing": { + "audio": "0.00000025", + "completion": "0.00000075", + "discount": 0, + "image": "0.000000125", + "input_audio_cache": "0.000000025", + "input_cache_read": "0.0000000125", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000075", + "prompt": "0.000000125", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite-preview", + "model_name": "Google: Gemini 3.1 Flash Lite Preview", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-preview-20260303", + "pricing": { + "audio": "0.0000005", + "completion": "0.0000015", + "discount": 0, + "image": "0.00000025", + "input_audio_cache": "0.00000005", + "input_cache_read": "0.000000025", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000015", + "prompt": "0.00000025", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.69931497440284, + "uptime_last_30m": 99.72320955910156, + "uptime_last_5m": 99.89232576350823 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.1-flash-lite-preview", + "model_name": "Google: Gemini 3.1 Flash Lite Preview", + "name": "Google AI Studio | google/gemini-3.1-flash-lite-preview-20260303", + "pricing": { + "audio": "0.0000009", + "completion": "0.0000027", + "discount": 0, + "image": "0.00000045", + "input_audio_cache": "0.00000009", + "input_cache_read": "0.000000045", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000027", + "prompt": "0.00000045", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.94301791374338, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3.1-flash-lite-preview", + "name": "Google: Gemini 3.1 Flash Lite Preview" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1779193800, + "description": "Gemini 3.5 Flash is Google's high-efficiency multimodal model, bringing near-Pro level coding and reasoning at Flash-tier cost and speed. It is highly optimized for coding proficiency and parallel agentic execution...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.0000015", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000075", + "input_audio_cache": "0.00000015", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.0000045", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 71.57676348547717, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google AI Studio | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.0000015", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000075", + "input_audio_cache": "0.00000015", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.0000045", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.89653388515262, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.000003", + "completion": "0.000009", + "discount": 0, + "image": "0.0000015", + "input_audio_cache": "0.0000003", + "input_cache_read": "0.00000015", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.000009", + "prompt": "0.0000015", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 98.850726777682, + "uptime_last_30m": 99.78591060321133, + "uptime_last_5m": 99.85911524373063 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google AI Studio | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.000003", + "completion": "0.000009", + "discount": 0, + "image": "0.0000015", + "input_audio_cache": "0.0000003", + "input_cache_read": "0.00000015", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.000009", + "prompt": "0.0000015", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.87073332634596, + "uptime_last_30m": 99.19168591224017, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.0000033", + "completion": "0.0000099", + "discount": 0, + "image": "0.00000165", + "input_audio_cache": "0.00000033", + "input_cache_read": "0.000000165", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000099", + "prompt": "0.00000165", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.0000054", + "completion": "0.0000162", + "discount": 0, + "image": "0.0000027", + "input_audio_cache": "0.00000054", + "input_cache_read": "0.00000027", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000162", + "prompt": "0.0000027", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.90625, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash", + "model_name": "Google: Gemini 3.5 Flash", + "name": "Google AI Studio | google/gemini-3.5-flash-20260519", + "pricing": { + "audio": "0.0000054", + "completion": "0.0000162", + "discount": 0, + "image": "0.0000027", + "input_audio_cache": "0.00000054", + "input_cache_read": "0.00000027", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000162", + "prompt": "0.0000027", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.9250936329588, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3.5-flash", + "name": "Google: Gemini 3.5 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1784646726, + "description": "Gemini 3.5 Flash Lite is a high-efficiency model from Google with upgraded agentic capabilities. It is suited for subagents that execute focused tasks within complex, multi-agent workflows.", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000015", + "completion": "0.00000125", + "discount": 0, + "image": "0.00000015", + "input_audio_cache": "0.000000015", + "input_cache_read": "0.000000015", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000125", + "prompt": "0.00000015", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 95.70581805952581, + "uptime_last_30m": 96.72619047619048, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google AI Studio | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000015", + "completion": "0.00000125", + "discount": 0, + "image": "0.00000015", + "input_audio_cache": "0.000000015", + "input_cache_read": "0.000000015", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000125", + "prompt": "0.00000015", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.98398472887278, + "uptime_last_30m": 99.9891058044274, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google AI Studio | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.0000003", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.00000003", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.97718350574888, + "uptime_last_30m": 99.99626921355022, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.0000003", + "completion": "0.0000025", + "discount": 0, + "image": "0.0000003", + "input_audio_cache": "0.00000003", + "input_cache_read": "0.00000003", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.0000025", + "prompt": "0.0000003", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.74386919250145, + "uptime_last_30m": 99.80845676501843, + "uptime_last_5m": 99.67206790123457 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000033", + "completion": "0.00000275", + "discount": 0, + "image": "0.00000033", + "input_audio_cache": "0.000000033", + "input_cache_read": "0.000000033", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.00000275", + "prompt": "0.00000033", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/eu", + "throughput_last_30m": null, + "uptime_last_1d": 99.93312527864468, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000033", + "completion": "0.00000275", + "discount": 0, + "image": "0.00000033", + "input_audio_cache": "0.000000033", + "input_cache_read": "0.000000033", + "input_cache_write": "0.0000000833333333333333", + "internal_reasoning": "0.00000275", + "prompt": "0.00000033", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us", + "throughput_last_30m": null, + "uptime_last_1d": 99.8529411764706, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000054", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000054", + "input_audio_cache": "0.000000054", + "input_cache_read": "0.000000054", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000045", + "prompt": "0.00000054", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.93058892626408, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.5-flash-lite", + "model_name": "Google: Gemini 3.5 Flash Lite", + "name": "Google AI Studio | google/gemini-3.5-flash-lite-20260721", + "pricing": { + "audio": "0.00000054", + "completion": "0.0000045", + "discount": 0, + "image": "0.00000054", + "input_audio_cache": "0.000000054", + "input_cache_read": "0.000000054", + "input_cache_write": "0.00000015", + "internal_reasoning": "0.0000045", + "prompt": "0.00000054", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.93908474834386, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "google/gemini-3.5-flash-lite", + "name": "Google: Gemini 3.5 Flash Lite" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1784646733, + "description": "Gemini 3.6 Flash is a high-efficiency model from Google for coding, agentic workflows, and web and app development. It is designed to produce polished outputs with fewer unnecessary edits and...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 85.31432934588953, + "uptime_last_30m": 92.03187250996015, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google AI Studio | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.97674688989653, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google AI Studio | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.75682493533147, + "uptime_last_30m": 99.88428158148506, + "uptime_last_5m": 99.88452655889145 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.75593238522453, + "uptime_last_30m": 99.9214248297538, + "uptime_last_5m": 99.91816693944354 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.98465787051242, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google AI Studio | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.9699789852897, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.6-flash", + "model_name": "Google: Gemini 3.6 Flash", + "name": "Google | google/gemini-3.6-flash-20260721", + "pricing": { + "audio": "0.000000825", + "completion": "0.000004125", + "discount": 0, + "image": "0.000000825", + "input_audio_cache": "0.0000000825", + "input_cache_read": "0.0000000825", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.000004125", + "prompt": "0.000000825", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3.6-flash", + "name": "Google: Gemini 3.6 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1786640581, + "description": "Gemini 3.7 Flash is a multimodal model from Google for fast agentic workflows, coding, and complex multi-step reasoning. It is designed for tasks that require responsive performance and reliable multi-step...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0.5, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 76.20306479075187, + "uptime_last_30m": 73.4536082474227, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google AI Studio | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0.5, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.8576497673029, + "uptime_last_30m": 99.8448543354594, + "uptime_last_5m": 99.83836206896551 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google AI Studio | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0.5, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.79697146101493, + "uptime_last_30m": 99.93153029784321, + "uptime_last_5m": 99.91319444444444 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0.5, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 98.93314936543464, + "uptime_last_30m": 99.56450734893849, + "uptime_last_5m": 99.43642127509686 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0.5, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.76019184652279, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.7-flash", + "model_name": "Google: Gemini 3.7 Flash", + "name": "Google AI Studio | google/gemini-3.7-flash-20260813", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0.5, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.9157101420532, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3.7-flash", + "name": "Google: Gemini 3.7 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Gemini" + }, + "created": 1788362056, + "description": "Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google AI Studio | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0.5, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.8601015925231, + "uptime_last_30m": 99.94076786387379, + "uptime_last_5m": 99.95948136142626 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.000000375", + "completion": "0.000001875", + "discount": 0.5, + "image": "0.000000375", + "input_audio_cache": "0.0000000375", + "input_cache_read": "0.0000000375", + "input_cache_write": "0.0000000208333333333333", + "internal_reasoning": "0.000001875", + "prompt": "0.000000375", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/flex", + "throughput_last_30m": null, + "uptime_last_1d": 69.64435732287875, + "uptime_last_30m": 87.87878787878788, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google AI Studio | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0.5, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio", + "throughput_last_30m": null, + "uptime_last_1d": 99.90677123108807, + "uptime_last_30m": 99.76670036550276, + "uptime_last_5m": 99.87699876998771 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.00000075", + "completion": "0.00000375", + "discount": 0.5, + "image": "0.00000075", + "input_audio_cache": "0.000000075", + "input_cache_read": "0.000000075", + "input_cache_write": "0.0000000416666666666667", + "internal_reasoning": "0.00000375", + "prompt": "0.00000075", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 97.55735004769164, + "uptime_last_30m": 98.51484429467813, + "uptime_last_5m": 98.66914401762952 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google AI Studio | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0.5, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google AI Studio", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-ai-studio/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.68312144466078, + "uptime_last_30m": 99.43438914027149, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "google/gemini-3.8-flash", + "model_name": "Google: Gemini 3.8 Flash", + "name": "Google | google/gemini-3.8-flash-20260902", + "pricing": { + "audio": "0.00000135", + "completion": "0.00000675", + "discount": 0.5, + "image": "0.00000135", + "input_audio_cache": "0.000000135", + "input_cache_read": "0.000000135", + "input_cache_write": "0.000000075", + "internal_reasoning": "0.00000675", + "prompt": "0.00000135", + "web_search": "0.014" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "response_format", + "stop", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.6804922785634, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "google/gemini-3.8-flash", + "name": "Google: Gemini 3.8 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1786302394, + "description": "Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "meta/muse-glimmer-30b", + "model_name": "Meta: Muse Glimmer 30B", + "name": "Phala | meta/muse-glimmer-30b-20260810", + "pricing": { + "completion": "0.0000011", + "discount": 0, + "input_cache_read": "0.00000004", + "prompt": "0.0000003" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.94800332778702, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "meta/muse-glimmer-30b", + "model_name": "Meta: Muse Glimmer 30B", + "name": "DeepInfra | meta/muse-glimmer-30b-20260810", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000004", + "prompt": "0.0000003" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.96926761548488, + "uptime_last_30m": 99.9748743718593, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "meta/muse-glimmer-30b", + "model_name": "Meta: Muse Glimmer 30B", + "name": "Fireworks | meta/muse-glimmer-30b-20260810", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.00000004", + "prompt": "0.00000035" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.987972095261, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "meta/muse-glimmer-30b", + "model_name": "Meta: Muse Glimmer 30B", + "name": "Together | meta/muse-glimmer-30b-20260810", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.00000004", + "prompt": "0.00000035" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.87482931269913, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "meta/muse-glimmer-30b", + "name": "Meta: Muse Glimmer 30B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video", + "file", + "audio" + ], + "instruct_type": null, + "modality": "text+image+file+audio+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1788378359, + "description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It is designed to keep track of information across extended tasks, work through...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "meta/muse-spark-1.3", + "model_name": "Meta: Muse Spark 1.3", + "name": "Meta | meta/muse-spark-1.3-20260902", + "pricing": { + "completion": "0.00000425", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000125", + "web_search": "0.0025" + }, + "provider_name": "Meta", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "repetition_penalty", + "top_k", + "temperature", + "top_p", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "meta", + "throughput_last_30m": null, + "uptime_last_1d": 99.99375359847045, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "meta/muse-spark-1.3", + "name": "Meta: Muse Spark 1.3" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1776699402, + "description": "Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding tasks across Python, Rust, and Go, and...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Inceptron | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000002835", + "discount": 0, + "input_cache_read": "0.0000000856", + "prompt": "0.000000471" + }, + "provider_name": "Inceptron", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "top_p", + "stop", + "frequency_penalty", + "logit_bias", + "parallel_tool_calls", + "presence_penalty", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inceptron/int4", + "throughput_last_30m": null, + "uptime_last_1d": 97.33943106969252, + "uptime_last_30m": 96.30566801619433, + "uptime_last_5m": 94.71544715447155 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Chutes | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.00000285", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.0000005" + }, + "provider_name": "Chutes", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "chutes/int4", + "throughput_last_30m": null, + "uptime_last_1d": 98.32819749938015, + "uptime_last_30m": 98.53826311263973, + "uptime_last_5m": 98.93992932862191 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "DigitalOcean | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "input_cache_read": "0.000000114", + "prompt": "0.00000057" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "structured_outputs", + "response_format", + "tool_choice", + "tools", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 97.69673704414588, + "uptime_last_30m": 99.90779160903642, + "uptime_last_5m": 99.52606635071089 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Decart | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000024696", + "discount": 0, + "input_cache_read": "0.0000000988", + "prompt": "0.0000005865" + }, + "provider_name": "Decart", + "quantization": "fp4", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "decart/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 86.37411984967915, + "uptime_last_30m": 77.4025069637883, + "uptime_last_5m": 65.80086580086581 + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 230400, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "StreamLake | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.00000252", + "discount": 0.37, + "input_cache_read": "0.0000001008", + "prompt": "0.0000005985" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "stop", + "top_p", + "temperature", + "max_tokens", + "tool_choice", + "tools", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.41590572706464, + "uptime_last_30m": 97.82707107288365, + "uptime_last_5m": 99.85632183908046 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "CoreWeave | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.00000341", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000065" + }, + "provider_name": "CoreWeave", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.81317001528609, + "uptime_last_30m": 99.78021978021978, + "uptime_last_5m": 99.66777408637874 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Crusoe | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000035", + "prompt": "0.0000007" + }, + "provider_name": "Crusoe", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "min_p", + "repetition_penalty", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "crusoe/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 97.64449247518412, + "uptime_last_30m": 99.64539007092199, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "DeepInfra | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000075" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tool_choice", + "tools", + "structured_outputs", + "response_format", + "logit_bias" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.64754524299045, + "uptime_last_30m": 99.33993399339934, + "uptime_last_5m": null + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Venice | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000075" + }, + "provider_name": "Venice", + "quantization": "int4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tool_choice", + "tools", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/int4", + "throughput_last_30m": null, + "uptime_last_1d": 73.13954233702967, + "uptime_last_30m": 80.61016949152543, + "uptime_last_5m": 83.52059925093633 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Parasail | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000075" + }, + "provider_name": "Parasail", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/int4", + "throughput_last_30m": null, + "uptime_last_1d": 99.89889801735825, + "uptime_last_30m": 99.85748218527316, + "uptime_last_5m": 99.84567901234568 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "SiliconFlow | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000034", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.00000077" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.76419618877094, + "uptime_last_30m": 99.14841849148418, + "uptime_last_5m": 97.28506787330316 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Novita | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000034", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.0000008" + }, + "provider_name": "Novita", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita", + "throughput_last_30m": null, + "uptime_last_1d": 99.47812191887941, + "uptime_last_30m": 99.77142857142857, + "uptime_last_5m": 99.67637540453075 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "GMICloud | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000036", + "discount": 0.1, + "input_cache_read": "0.000000144", + "prompt": "0.000000855" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.53048397234444, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Baidu | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "Baidu", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "frequency_penalty", + "presence_penalty", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.98705103510788, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "AtlasCloud | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "AtlasCloud", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "tool_choice", + "tools", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/int4", + "throughput_last_30m": null, + "uptime_last_1d": 98.77813139349853, + "uptime_last_30m": 98.55072463768117, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Cloudflare | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "min_p", + "stop", + "logit_bias", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.93317245219974, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Moonshot AI | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "Moonshot AI", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "stop", + "frequency_penalty", + "presence_penalty", + "structured_outputs", + "response_format", + "tool_choice", + "tools" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "moonshotai/int4", + "throughput_last_30m": null, + "uptime_last_1d": 99.55261088275883, + "uptime_last_30m": 99.50298210735586, + "uptime_last_5m": 99.079754601227 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Phala | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.0000046", + "discount": 0, + "input_cache_read": "0.00000037", + "prompt": "0.00000109" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "structured_outputs", + "response_format", + "tool_choice", + "tools" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.29348378570702, + "uptime_last_30m": 99.7624703087886, + "uptime_last_5m": 100 + }, + { + "context_length": 262000, + "latency_last_30m": null, + "max_completion_tokens": 235800, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "BaseTen | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 97.22386997271325, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "Fireworks | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 0, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 262000, + "latency_last_30m": null, + "max_completion_tokens": 235800, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k2.6", + "model_name": "MoonshotAI: Kimi K2.6", + "name": "BaseTen | moonshotai/kimi-k2.6-20260420", + "pricing": { + "completion": "0.000004", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "moonshotai/kimi-k2.6", + "name": "MoonshotAI: Kimi K2.6" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1784215858, + "description": "Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "InferenceNet | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001095", + "discount": 0, + "input_cache_read": "0.00000023", + "prompt": "0.0000021" + }, + "provider_name": "InferenceNet", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inference-net/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.04227219711652, + "uptime_last_30m": 95.82336045955002, + "uptime_last_5m": 94.01373895976447 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Relace | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001095", + "discount": 0, + "input_cache_read": "0.00000023", + "prompt": "0.0000021" + }, + "provider_name": "Relace", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "relace/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.87092770477129, + "uptime_last_30m": 99.93888775718068, + "uptime_last_5m": 99.92088607594937 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Sail Research | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000010758", + "discount": 0, + "input_cache_read": "0.000000244", + "prompt": "0.000002144" + }, + "provider_name": "Sail Research", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "max_tokens", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.69528884902287, + "uptime_last_30m": 99.26873857404021, + "uptime_last_5m": 99.65753424657534 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Phala | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000012", + "discount": 0.2, + "input_cache_read": "0.00000024", + "prompt": "0.0000024" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.81455362273415, + "uptime_last_30m": 98.64176570458405, + "uptime_last_5m": 88.09523809523809 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Wafer | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001095", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.0000025" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer", + "throughput_last_30m": null, + "uptime_last_1d": 99.85483402690409, + "uptime_last_30m": 91.12903225806451, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Morph | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000014", + "discount": 0, + "input_cache_read": "0.00000029", + "prompt": "0.0000025" + }, + "provider_name": "Morph", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "morph/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.69838940172261, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Makora | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001275", + "discount": 0, + "input_cache_read": "0.00000025599999", + "prompt": "0.00000255" + }, + "provider_name": "Makora", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "max_tokens", + "seed", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "makora", + "throughput_last_30m": null, + "uptime_last_1d": 97.91822674853977, + "uptime_last_30m": 90.06211180124224, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "DigitalOcean | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001295", + "discount": 0, + "input_cache_read": "0.000000255", + "prompt": "0.00000255" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 99.30996216302366, + "uptime_last_30m": 99.80292792792793, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "DeepInfra | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001425", + "discount": 0, + "input_cache_read": "0.000000285", + "prompt": "0.00000285" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 97.93592976557393, + "uptime_last_30m": 99.40564635958395, + "uptime_last_5m": 99.04761904761905 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Chutes | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Chutes", + "quantization": "mxfp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "chutes/mxfp4", + "throughput_last_30m": null, + "uptime_last_1d": 97.12998102466793, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Parasail | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Parasail", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.19240949489654, + "uptime_last_30m": 99.56822107081175, + "uptime_last_5m": 98.9010989010989 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Modal | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Modal", + "quantization": "mxfp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "modal/mxfp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.64235624123422, + "uptime_last_30m": 99.72640218878249, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Together | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.53068395320594, + "uptime_last_30m": 99.078125, + "uptime_last_5m": 99.6067415730337 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Fireworks | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.90984609440402, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "BaseTen | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.79443072468705, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Moonshot AI | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.000003" + }, + "provider_name": "Moonshot AI", + "quantization": "mxfp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "stop", + "frequency_penalty", + "presence_penalty", + "structured_outputs", + "response_format", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "moonshotai/mxfp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.7809471573224, + "uptime_last_30m": 99.97940055618498, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Fireworks | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.0000165", + "discount": 0, + "input_cache_read": "0.00000033", + "prompt": "0.0000033" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks/us", + "throughput_last_30m": null, + "uptime_last_1d": 99.0821795760207, + "uptime_last_30m": 99.85754985754986, + "uptime_last_5m": 99.66442953020133 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Alibaba | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.00001725", + "discount": 0, + "input_cache_read": "0.000000345", + "prompt": "0.00000345" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "seed", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.09021155321715, + "uptime_last_30m": 98, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Fireworks | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.0000225", + "discount": 0, + "input_cache_read": "0.00000045", + "prompt": "0.0000045" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.20883917610148, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "moonshotai/kimi-k3", + "model_name": "MoonshotAI: Kimi K3", + "name": "Morph | moonshotai/kimi-k3-20260715", + "pricing": { + "completion": "0.0000225", + "discount": 0, + "input_cache_read": "0.0000006", + "prompt": "0.000006" + }, + "provider_name": "Morph", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "morph/fast", + "throughput_last_30m": null, + "uptime_last_1d": 97.88491964360063, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "moonshotai/kimi-k3", + "name": "MoonshotAI: Kimi K3" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1706140800, + "description": "GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks.\n\nTraining data up to Sep 2021.", + "endpoints": [ + { + "context_length": 4095, + "latency_last_30m": null, + "max_completion_tokens": 3685, + "max_prompt_tokens": null, + "model_id": "openai/gpt-3.5-turbo-0613", + "model_name": "OpenAI: GPT-3.5 Turbo (older v0613)", + "name": "Azure | openai/gpt-3.5-turbo-0613", + "pricing": { + "completion": "0.000002", + "discount": 0, + "prompt": "0.000001" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-3.5-turbo-0613", + "name": "OpenAI: GPT-3.5 Turbo (older v0613)" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1693180800, + "description": "This model offers four times the context length of gpt-3.5-turbo, allowing it to support approximately 20 pages of text in a single request at a higher cost. Training data: up...", + "endpoints": [ + { + "context_length": 16385, + "latency_last_30m": null, + "max_completion_tokens": 4096, + "max_prompt_tokens": null, + "model_id": "openai/gpt-3.5-turbo-16k", + "model_name": "OpenAI: GPT-3.5 Turbo 16k", + "name": "Azure | openai/gpt-3.5-turbo-16k", + "pricing": { + "completion": "0.000004", + "discount": 0, + "prompt": "0.000003" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 16385, + "latency_last_30m": null, + "max_completion_tokens": 4096, + "max_prompt_tokens": null, + "model_id": "openai/gpt-3.5-turbo-16k", + "model_name": "OpenAI: GPT-3.5 Turbo 16k", + "name": "OpenAI | openai/gpt-3.5-turbo-16k", + "pricing": { + "completion": "0.000004", + "discount": 0, + "prompt": "0.000003" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-3.5-turbo-16k", + "name": "OpenAI: GPT-3.5 Turbo 16k" + }, + { + "architecture": { + "input_modalities": [ + "image", + "text", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1744651385, + "description": "GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning. It supports a 1 million token context window and outperforms GPT-4o and...", + "endpoints": [ + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1", + "model_name": "OpenAI: GPT-4.1", + "name": "Azure | openai/gpt-4.1-2025-04-14", + "pricing": { + "completion": "0.000008", + "discount": 0, + "input_cache_read": "0.0000005", + "prompt": "0.000002", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.97266816272513, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1", + "model_name": "OpenAI: GPT-4.1", + "name": "OpenAI | openai/gpt-4.1-2025-04-14", + "pricing": { + "completion": "0.000008", + "discount": 0, + "input_cache_read": "0.0000005", + "prompt": "0.000002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.97033818099801, + "uptime_last_30m": 99.97643510227165, + "uptime_last_5m": 99.94345490528697 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1", + "model_name": "OpenAI: GPT-4.1", + "name": "Azure | openai/gpt-4.1-2025-04-14", + "pricing": { + "completion": "0.0000088", + "discount": 0, + "input_cache_read": "0.00000055", + "prompt": "0.0000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-4.1", + "name": "OpenAI: GPT-4.1" + }, + { + "architecture": { + "input_modalities": [ + "image", + "text", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1744651381, + "description": "GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.1% on hard...", + "endpoints": [ + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-mini", + "model_name": "OpenAI: GPT-4.1 Mini", + "name": "Azure | openai/gpt-4.1-mini-2025-04-14", + "pricing": { + "completion": "0.0000016", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.0000004", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.96720636839466, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-mini", + "model_name": "OpenAI: GPT-4.1 Mini", + "name": "OpenAI | openai/gpt-4.1-mini-2025-04-14", + "pricing": { + "completion": "0.0000016", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.0000004", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.93112635620022, + "uptime_last_30m": 99.94685510141818, + "uptime_last_5m": 99.97336352134248 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-mini", + "model_name": "OpenAI: GPT-4.1 Mini", + "name": "Azure | openai/gpt-4.1-mini-2025-04-14", + "pricing": { + "completion": "0.00000176", + "discount": 0, + "input_cache_read": "0.00000011", + "prompt": "0.00000044", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-4.1-mini", + "name": "OpenAI: GPT-4.1 Mini" + }, + { + "architecture": { + "input_modalities": [ + "image", + "text", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1744651369, + "description": "For tasks that demand low latency, GPT\u20114.1 nano is the fastest and cheapest model in the GPT-4.1 series. It delivers exceptional performance at a small size with its 1 million...", + "endpoints": [ + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-nano", + "model_name": "OpenAI: GPT-4.1 Nano", + "name": "Azure | openai/gpt-4.1-nano-2025-04-14", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000001", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.91645467431404, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-nano", + "model_name": "OpenAI: GPT-4.1 Nano", + "name": "OpenAI | openai/gpt-4.1-nano-2025-04-14", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.000000025", + "prompt": "0.0000001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "temperature", + "top_p" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.96280114824496, + "uptime_last_30m": 99.99066961092278, + "uptime_last_5m": 99.98529988239906 + }, + { + "context_length": 1047576, + "latency_last_30m": null, + "max_completion_tokens": 942818, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4.1-nano", + "model_name": "OpenAI: GPT-4.1 Nano", + "name": "Azure | openai/gpt-4.1-nano-2025-04-14", + "pricing": { + "completion": "0.00000044", + "discount": 0, + "input_cache_read": "0.000000033", + "prompt": "0.00000011", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "seed", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "temperature", + "top_p" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-4.1-nano", + "name": "OpenAI: GPT-4.1 Nano" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1715558400, + "description": "GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o", + "model_name": "OpenAI: GPT-4o", + "name": "Azure | openai/gpt-4o", + "pricing": { + "completion": "0.00001", + "discount": 0, + "prompt": "0.0000025" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.9243451283848, + "uptime_last_30m": 99.93881921076782, + "uptime_last_5m": 100 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o", + "model_name": "OpenAI: GPT-4o", + "name": "OpenAI | openai/gpt-4o", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.00000125", + "prompt": "0.0000025" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "logit_bias", + "logprobs", + "top_logprobs", + "prediction", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.87503943227308, + "uptime_last_30m": 99.50179044060408, + "uptime_last_5m": 99.73544973544973 + } + ], + "id": "openai/gpt-4o", + "name": "OpenAI: GPT-4o" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1722902400, + "description": "The 2024-08-06 version of GPT-4o offers improved performance in structured outputs, with the ability to supply a JSON schema in the respone_format. Read more [here](https://openai.com/index/introducing-structured-outputs-in-the-api/). GPT-4o (\"o\" for \"omni\") is...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-2024-08-06", + "model_name": "OpenAI: GPT-4o (2024-08-06)", + "name": "Azure | openai/gpt-4o-2024-08-06", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.00000125", + "prompt": "0.0000025" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.97736418511066, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-2024-08-06", + "model_name": "OpenAI: GPT-4o (2024-08-06)", + "name": "OpenAI | openai/gpt-4o-2024-08-06", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.00000125", + "prompt": "0.0000025" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "logit_bias", + "logprobs", + "top_logprobs", + "prediction", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.93595507001835, + "uptime_last_30m": 99.56268221574344, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-4o-2024-08-06", + "name": "OpenAI: GPT-4o (2024-08-06)" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1732127594, + "description": "The 2024-11-20 version of GPT-4o offers a leveled-up creative writing ability with more natural, engaging, and tailored writing to improve relevance & readability. It\u2019s also better at working with uploaded...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-2024-11-20", + "model_name": "OpenAI: GPT-4o (2024-11-20)", + "name": "OpenAI | openai/gpt-4o-2024-11-20", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.00000125", + "prompt": "0.0000025" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "logit_bias", + "logprobs", + "top_logprobs", + "prediction", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-4o-2024-11-20", + "name": "OpenAI: GPT-4o (2024-11-20)" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1721260800, + "description": "GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-mini", + "model_name": "OpenAI: GPT-4o-mini", + "name": "Azure | openai/gpt-4o-mini", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000015" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 98.39699719555614, + "uptime_last_30m": 99.86744343127322, + "uptime_last_5m": 99.81925533196771 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-mini", + "model_name": "OpenAI: GPT-4o-mini", + "name": "OpenAI | openai/gpt-4o-mini", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000015" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "logit_bias", + "logprobs", + "top_logprobs", + "prediction", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.94346782001838, + "uptime_last_30m": 99.99429454958322, + "uptime_last_5m": 99.99628639334522 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-mini", + "model_name": "OpenAI: GPT-4o-mini", + "name": "Azure | openai/gpt-4o-mini", + "pricing": { + "completion": "0.00000066", + "discount": 0, + "input_cache_read": "0.0000000825", + "prompt": "0.000000165" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "seed", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-4o-mini", + "name": "OpenAI: GPT-4o-mini" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1721260800, + "description": "GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-4o-mini-2024-07-18", + "model_name": "OpenAI: GPT-4o-mini (2024-07-18)", + "name": "OpenAI | openai/gpt-4o-mini-2024-07-18", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000015" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "web_search_options", + "logit_bias", + "logprobs", + "top_logprobs", + "prediction", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-4o-mini-2024-07-18", + "name": "OpenAI: GPT-4o-mini (2024-07-18)" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1754587413, + "description": "GPT-5 is OpenAI\u2019s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5", + "model_name": "OpenAI: GPT-5", + "name": "Azure | openai/gpt-5-2025-08-07", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.000000125", + "prompt": "0.00000125", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.98832048586779, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5", + "model_name": "OpenAI: GPT-5", + "name": "OpenAI | openai/gpt-5-2025-08-07", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.000000125", + "prompt": "0.00000125", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/default", + "throughput_last_30m": null, + "uptime_last_1d": 99.9797096490565, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5", + "model_name": "OpenAI: GPT-5", + "name": "Azure | openai/gpt-5-2025-08-07", + "pricing": { + "completion": "0.000011", + "discount": 0, + "input_cache_read": "0.0000001375", + "prompt": "0.000001375", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5", + "name": "OpenAI: GPT-5" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1754587407, + "description": "GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5-mini", + "model_name": "OpenAI: GPT-5 Mini", + "name": "OpenAI | openai/gpt-5-mini-2025-08-07", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.0000000125", + "prompt": "0.000000125", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 360000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5-mini", + "model_name": "OpenAI: GPT-5 Mini", + "name": "Azure | openai/gpt-5-mini-2025-08-07", + "pricing": { + "completion": "0.000002", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000025", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99795496840426, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5-mini", + "model_name": "OpenAI: GPT-5 Mini", + "name": "OpenAI | openai/gpt-5-mini-2025-08-07", + "pricing": { + "completion": "0.000002", + "discount": 0, + "input_cache_read": "0.000000025", + "prompt": "0.00000025", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.97328508778891, + "uptime_last_30m": 99.97754394969844, + "uptime_last_5m": 99.93983876789797 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 360000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5-mini", + "model_name": "OpenAI: GPT-5 Mini", + "name": "Azure | openai/gpt-5-mini-2025-08-07", + "pricing": { + "completion": "0.0000022", + "discount": 0, + "input_cache_read": "0.000000033", + "prompt": "0.000000275", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5-mini", + "name": "OpenAI: GPT-5 Mini" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1754587402, + "description": "GPT-5-Nano is the smallest and fastest variant in the GPT-5 system, optimized for developer tools, rapid interactions, and ultra-low latency environments. While limited in reasoning depth compared to its larger...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5-nano", + "model_name": "OpenAI: GPT-5 Nano", + "name": "OpenAI | openai/gpt-5-nano-2025-08-07", + "pricing": { + "completion": "0.0000002", + "discount": 0, + "input_cache_read": "0.0000000025", + "prompt": "0.000000025", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 360000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5-nano", + "model_name": "OpenAI: GPT-5 Nano", + "name": "Azure | openai/gpt-5-nano-2025-08-07", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.00000001", + "prompt": "0.00000005", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.98822143698469, + "uptime_last_30m": 99.98315363881402, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5-nano", + "model_name": "OpenAI: GPT-5 Nano", + "name": "OpenAI | openai/gpt-5-nano-2025-08-07", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.000000005", + "prompt": "0.00000005", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.97633949107086, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 360000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5-nano", + "model_name": "OpenAI: GPT-5 Nano", + "name": "Azure | openai/gpt-5-nano-2025-08-07", + "pricing": { + "completion": "0.00000044", + "discount": 0, + "input_cache_read": "0.000000011", + "prompt": "0.000000055", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5-nano", + "name": "OpenAI: GPT-5 Nano" + }, + { + "architecture": { + "input_modalities": [ + "image", + "text", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1763060305, + "description": "GPT-5.1 is the latest frontier-grade model in the GPT-5 series, offering stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style compared to GPT-5. It uses adaptive reasoning...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.1", + "model_name": "OpenAI: GPT-5.1", + "name": "OpenAI | openai/gpt-5.1-20251113", + "pricing": { + "completion": "0.000005", + "discount": 0, + "input_cache_read": "0.0000000625", + "prompt": "0.000000625", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.88412514484357, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.1", + "model_name": "OpenAI: GPT-5.1", + "name": "Azure | openai/gpt-5.1-20251113", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.00000013", + "prompt": "0.00000125", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99710203726781, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.1", + "model_name": "OpenAI: GPT-5.1", + "name": "OpenAI | openai/gpt-5.1-20251113", + "pricing": { + "completion": "0.00001", + "discount": 0, + "input_cache_read": "0.000000125", + "prompt": "0.00000125", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.97425152158581, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.1", + "model_name": "OpenAI: GPT-5.1", + "name": "Azure | openai/gpt-5.1-20251113", + "pricing": { + "completion": "0.000011", + "discount": 0, + "input_cache_read": "0.000000143", + "prompt": "0.000001375", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/swedencentral", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.1", + "model_name": "OpenAI: GPT-5.1", + "name": "OpenAI | openai/gpt-5.1-20251113", + "pricing": { + "completion": "0.00002", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.0000025", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.1", + "name": "OpenAI: GPT-5.1" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1765389775, + "description": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.2", + "model_name": "OpenAI: GPT-5.2", + "name": "OpenAI | openai/gpt-5.2-20251211", + "pricing": { + "completion": "0.000007", + "discount": 0, + "input_cache_read": "0.0000000875", + "prompt": "0.000000875", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.2", + "model_name": "OpenAI: GPT-5.2", + "name": "Azure | openai/gpt-5.2-20251211", + "pricing": { + "completion": "0.000014", + "discount": 0, + "input_cache_read": "0.000000175", + "prompt": "0.00000175", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99522684422806, + "uptime_last_30m": 99.20634920634922, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.2", + "model_name": "OpenAI: GPT-5.2", + "name": "OpenAI | openai/gpt-5.2-20251211", + "pricing": { + "completion": "0.000014", + "discount": 0, + "input_cache_read": "0.000000175", + "prompt": "0.00000175", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.91117616249453, + "uptime_last_30m": 99.96385977593062, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 272000, + "model_id": "openai/gpt-5.2", + "model_name": "OpenAI: GPT-5.2", + "name": "OpenAI | openai/gpt-5.2-20251211", + "pricing": { + "completion": "0.000028", + "discount": 0, + "input_cache_read": "0.00000035", + "prompt": "0.0000035", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.2", + "name": "OpenAI: GPT-5.2" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1772734352, + "description": "GPT-5.4 is OpenAI\u2019s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...", + "endpoints": [ + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "OpenAI | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.0000075", + "discount": 0, + "input_cache_read": "0.000000125", + "overrides": [ + { + "completion": "0.00001125", + "input_cache_read": "0.00000025", + "min_prompt_tokens": 272000, + "prompt": "0.0000025" + } + ], + "prompt": "0.00000125", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 48.614424665076164, + "uptime_last_30m": 56.16438356164384, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "Azure | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.00000025", + "overrides": [ + { + "completion": "0.0000225", + "input_cache_read": "0.0000005", + "min_prompt_tokens": 272000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.9454411245547, + "uptime_last_30m": 99.86149584487535, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "OpenAI | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.000015", + "discount": 0, + "input_cache_read": "0.00000025", + "overrides": [ + { + "completion": "0.0000225", + "input_cache_read": "0.0000005", + "min_prompt_tokens": 272000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.94598535841328, + "uptime_last_30m": 99.96911037891269, + "uptime_last_5m": 99.92418498862776 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "Azure | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.0000165", + "discount": 0, + "input_cache_read": "0.000000275", + "overrides": [ + { + "completion": "0.00002475", + "input_cache_read": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000055" + } + ], + "prompt": "0.00000275", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "Azure | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.0000165", + "discount": 0, + "input_cache_read": "0.000000275", + "overrides": [ + { + "completion": "0.00002475", + "input_cache_read": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000055" + } + ], + "prompt": "0.00000275", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/eu", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "Amazon Bedrock | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.0000165", + "discount": 0, + "input_cache_read": "0.000000275", + "overrides": [ + { + "completion": "0.00002475", + "input_cache_read": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000055" + } + ], + "prompt": "0.00000275", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/us-east-1", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.4", + "model_name": "OpenAI: GPT-5.4", + "name": "OpenAI | openai/gpt-5.4-20260305", + "pricing": { + "completion": "0.00003", + "discount": 0, + "input_cache_read": "0.0000005", + "overrides": [ + { + "completion": "0.000045", + "input_cache_read": "0.000001", + "min_prompt_tokens": 272000, + "prompt": "0.00001" + } + ], + "prompt": "0.000005", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.4", + "name": "OpenAI: GPT-5.4" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1773748178, + "description": "GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-mini", + "model_name": "OpenAI: GPT-5.4 Mini", + "name": "OpenAI | openai/gpt-5.4-mini-20260317", + "pricing": { + "completion": "0.00000225", + "discount": 0, + "input_cache_read": "0.0000000375", + "prompt": "0.000000375", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 97.99155773420479, + "uptime_last_30m": null, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-mini", + "model_name": "OpenAI: GPT-5.4 Mini", + "name": "Azure | openai/gpt-5.4-mini-20260317", + "pricing": { + "completion": "0.0000045", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000075", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99079290770598, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-mini", + "model_name": "OpenAI: GPT-5.4 Mini", + "name": "OpenAI | openai/gpt-5.4-mini-20260317", + "pricing": { + "completion": "0.0000045", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000075", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.95526220552502, + "uptime_last_30m": 99.98510264279116, + "uptime_last_5m": 99.94242947610823 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-mini", + "model_name": "OpenAI: GPT-5.4 Mini", + "name": "Azure | openai/gpt-5.4-mini-20260317", + "pricing": { + "completion": "0.00000495", + "discount": 0, + "input_cache_read": "0.0000000825", + "prompt": "0.000000825", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-mini", + "model_name": "OpenAI: GPT-5.4 Mini", + "name": "OpenAI | openai/gpt-5.4-mini-20260317", + "pricing": { + "completion": "0.000009", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.0000015", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.96334982591168, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.4-mini", + "name": "OpenAI: GPT-5.4 Mini" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1773748187, + "description": "GPT-5.4 nano is the most lightweight and cost-efficient variant of the GPT-5.4 family, optimized for speed-critical and high-volume tasks. It supports text and image inputs and is designed for low-latency...", + "endpoints": [ + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-nano", + "model_name": "OpenAI: GPT-5.4 Nano", + "name": "OpenAI | openai/gpt-5.4-nano-20260317", + "pricing": { + "completion": "0.000000625", + "discount": 0, + "input_cache_read": "0.00000001", + "prompt": "0.0000001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-nano", + "model_name": "OpenAI: GPT-5.4 Nano", + "name": "Azure | openai/gpt-5.4-nano-20260317", + "pricing": { + "completion": "0.00000125", + "discount": 0, + "input_cache_read": "0.00000002", + "prompt": "0.0000002", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99363651404603, + "uptime_last_30m": 99.99136740331491, + "uptime_last_5m": 99.94722955145119 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-nano", + "model_name": "OpenAI: GPT-5.4 Nano", + "name": "OpenAI | openai/gpt-5.4-nano-20260317", + "pricing": { + "completion": "0.00000125", + "discount": 0, + "input_cache_read": "0.00000002", + "prompt": "0.0000002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.9841333065326, + "uptime_last_30m": 99.99315334231007, + "uptime_last_5m": 100 + }, + { + "context_length": 400000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.4-nano", + "model_name": "OpenAI: GPT-5.4 Nano", + "name": "Azure | openai/gpt-5.4-nano-20260317", + "pricing": { + "completion": "0.000001375", + "discount": 0, + "input_cache_read": "0.000000022", + "prompt": "0.00000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.4-nano", + "name": "OpenAI: GPT-5.4 Nano" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1783590864, + "description": "GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...", + "endpoints": [ + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "OpenAI | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.00000001", + "input_cache_write": "0.000000125", + "overrides": [ + { + "completion": "0.0000009", + "input_cache_read": "0.00000002", + "input_cache_write": "0.00000025", + "min_prompt_tokens": 272000, + "prompt": "0.0000002" + } + ], + "prompt": "0.0000001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.94307072216482, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "Azure | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000002", + "input_cache_write": "0.00000025", + "overrides": [ + { + "completion": "0.0000018", + "input_cache_read": "0.00000004", + "input_cache_write": "0.0000005", + "min_prompt_tokens": 272000, + "prompt": "0.0000004" + } + ], + "prompt": "0.0000002", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 89.2375911535796, + "uptime_last_30m": 97.27737596167135, + "uptime_last_5m": 92.6924326079896 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "OpenAI | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "input_cache_read": "0.00000002", + "input_cache_write": "0.00000025", + "overrides": [ + { + "completion": "0.0000018", + "input_cache_read": "0.00000004", + "input_cache_write": "0.0000005", + "min_prompt_tokens": 272000, + "prompt": "0.0000004" + } + ], + "prompt": "0.0000002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.99005784292184, + "uptime_last_30m": 99.98960699597642, + "uptime_last_5m": 99.9799802304776 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "Azure | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000022", + "input_cache_write": "0.000000275", + "overrides": [ + { + "completion": "0.00000198", + "input_cache_read": "0.000000044", + "input_cache_write": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.00000044" + } + ], + "prompt": "0.00000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": 99.99503105590063, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "Amazon Bedrock | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000022", + "input_cache_write": "0.000000275", + "overrides": [ + { + "completion": "0.00000198", + "input_cache_read": "0.000000044", + "input_cache_write": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.00000044" + } + ], + "prompt": "0.00000022", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/us-east-1", + "throughput_last_30m": null, + "uptime_last_1d": 99.99202653568922, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "Azure | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.00000132", + "discount": 0, + "input_cache_read": "0.000000022", + "input_cache_write": "0.000000275", + "overrides": [ + { + "completion": "0.00000198", + "input_cache_read": "0.000000044", + "input_cache_write": "0.00000055", + "min_prompt_tokens": 272000, + "prompt": "0.00000044" + } + ], + "prompt": "0.00000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/eu", + "throughput_last_30m": null, + "uptime_last_1d": 98.21728160386894, + "uptime_last_30m": 95.3660870952305, + "uptime_last_5m": 95.40575354229283 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-luna", + "model_name": "OpenAI: GPT-5.6 Luna", + "name": "OpenAI | openai/gpt-5.6-luna-20260709", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "input_cache_read": "0.00000004", + "input_cache_write": "0.0000005", + "overrides": [ + { + "completion": "0.0000036", + "input_cache_read": "0.00000008", + "input_cache_write": "0.000001", + "min_prompt_tokens": 272000, + "prompt": "0.0000008" + } + ], + "prompt": "0.0000004", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.9914811276553, + "uptime_last_30m": 99.98958767180342, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-5.6-luna", + "name": "OpenAI: GPT-5.6 Luna" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1783590850, + "description": "GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks...", + "endpoints": [ + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "OpenAI | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.000005", + "discount": 0.5, + "input_cache_read": "0.0000001", + "input_cache_write": "0.00000125", + "overrides": [ + { + "completion": "0.0000075", + "input_cache_read": "0.0000002", + "input_cache_write": "0.0000025", + "min_prompt_tokens": 272000, + "prompt": "0.000002" + } + ], + "prompt": "0.000001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 84.15062126529452, + "uptime_last_30m": 47.9328165374677, + "uptime_last_5m": 98.91304347826086 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "OpenAI | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.00001", + "discount": 0.5, + "input_cache_read": "0.0000002", + "input_cache_write": "0.0000025", + "overrides": [ + { + "completion": "0.000015", + "input_cache_read": "0.0000004", + "input_cache_write": "0.000005", + "min_prompt_tokens": 272000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.93752497274579, + "uptime_last_30m": 99.97973575423524, + "uptime_last_5m": 99.9452140565078 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "OpenAI | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.00002", + "discount": 0.5, + "input_cache_read": "0.0000004", + "input_cache_write": "0.000005", + "overrides": [ + { + "completion": "0.00003", + "input_cache_read": "0.0000008", + "input_cache_write": "0.00001", + "min_prompt_tokens": 272000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.93444253549146, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "Amazon Bedrock | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.000022", + "discount": 0, + "input_cache_read": "0.00000044", + "input_cache_write": "0.0000055", + "overrides": [ + { + "completion": "0.000033", + "input_cache_read": "0.00000088", + "input_cache_write": "0.000011", + "min_prompt_tokens": 272000, + "prompt": "0.0000088" + } + ], + "prompt": "0.0000044", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/us-east-1", + "throughput_last_30m": null, + "uptime_last_1d": 99.9715424018213, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "Azure | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.00003", + "discount": 0, + "input_cache_read": "0.0000005", + "input_cache_write": "0.00000625", + "overrides": [ + { + "completion": "0.000045", + "input_cache_read": "0.000001", + "input_cache_write": "0.0000125", + "min_prompt_tokens": 272000, + "prompt": "0.00001" + } + ], + "prompt": "0.000005", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.98722618771889, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "Azure | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.000033", + "discount": 0, + "input_cache_read": "0.00000055", + "input_cache_write": "0.000006875", + "overrides": [ + { + "completion": "0.0000495", + "input_cache_read": "0.0000011", + "input_cache_write": "0.00001375", + "min_prompt_tokens": 272000, + "prompt": "0.000011" + } + ], + "prompt": "0.0000055", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-sol", + "model_name": "OpenAI: GPT-5.6 Sol", + "name": "Azure | openai/gpt-5.6-sol-20260709", + "pricing": { + "completion": "0.000033", + "discount": 0, + "input_cache_read": "0.00000055", + "input_cache_write": "0.000006875", + "overrides": [ + { + "completion": "0.0000495", + "input_cache_read": "0.0000011", + "input_cache_write": "0.00001375", + "min_prompt_tokens": 272000, + "prompt": "0.000011" + } + ], + "prompt": "0.0000055", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/eu", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.6-sol", + "name": "OpenAI: GPT-5.6 Sol" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1783590857, + "description": "GPT-5.6 Terra is a balanced model in OpenAI's GPT-5.6 series, positioned between the flagship Sol tier and the cost-efficient Luna tier. It is suited for everyday coding, reasoning, and agentic...", + "endpoints": [ + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "OpenAI | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000001", + "input_cache_write": "0.00000125", + "overrides": [ + { + "completion": "0.000009", + "input_cache_read": "0.0000002", + "input_cache_write": "0.0000025", + "min_prompt_tokens": 272000, + "prompt": "0.000002" + } + ], + "prompt": "0.000001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.85897910116516, + "uptime_last_30m": 99.11971830985915, + "uptime_last_5m": 95.94594594594594 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "Azure | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.0000002", + "input_cache_write": "0.0000025", + "overrides": [ + { + "completion": "0.000018", + "input_cache_read": "0.0000004", + "input_cache_write": "0.000005", + "min_prompt_tokens": 272000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.99671819106692, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "OpenAI | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.0000002", + "input_cache_write": "0.0000025", + "overrides": [ + { + "completion": "0.000018", + "input_cache_read": "0.0000004", + "input_cache_write": "0.000005", + "min_prompt_tokens": 272000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.97684021887117, + "uptime_last_30m": 99.98698449295304, + "uptime_last_5m": 99.96905461859818 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "Azure | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.0000132", + "discount": 0, + "input_cache_read": "0.00000022", + "input_cache_write": "0.00000275", + "overrides": [ + { + "completion": "0.0000198", + "input_cache_read": "0.00000044", + "input_cache_write": "0.0000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000044" + } + ], + "prompt": "0.0000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "Amazon Bedrock | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.0000132", + "discount": 0, + "input_cache_read": "0.00000022", + "input_cache_write": "0.00000275", + "overrides": [ + { + "completion": "0.0000198", + "input_cache_read": "0.00000044", + "input_cache_write": "0.0000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000044" + } + ], + "prompt": "0.0000022", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/us-east-1", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "Azure | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.0000132", + "discount": 0, + "input_cache_read": "0.00000022", + "input_cache_write": "0.00000275", + "overrides": [ + { + "completion": "0.0000198", + "input_cache_read": "0.00000044", + "input_cache_write": "0.0000055", + "min_prompt_tokens": 272000, + "prompt": "0.0000044" + } + ], + "prompt": "0.0000022", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/eu", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-5.6-terra", + "model_name": "OpenAI: GPT-5.6 Terra", + "name": "OpenAI | openai/gpt-5.6-terra-20260709", + "pricing": { + "completion": "0.000024", + "discount": 0, + "input_cache_read": "0.0000004", + "input_cache_write": "0.000005", + "overrides": [ + { + "completion": "0.000036", + "input_cache_read": "0.0000008", + "input_cache_write": "0.00001", + "min_prompt_tokens": 272000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "openai/gpt-5.6-terra", + "name": "OpenAI: GPT-5.6 Terra" + }, + { + "architecture": { + "input_modalities": [ + "file", + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1788552838, + "description": "GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...", + "endpoints": [ + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-6-astra", + "model_name": "OpenAI: GPT-6 Astra", + "name": "OpenAI | openai/gpt-6-astra-20260903", + "pricing": { + "completion": "0.000025", + "discount": 0, + "input_cache_read": "0.0000005", + "input_cache_write": "0.00000625", + "overrides": [ + { + "completion": "0.0000375", + "input_cache_read": "0.000001", + "input_cache_write": "0.0000125", + "min_prompt_tokens": 272000, + "prompt": "0.00001" + } + ], + "prompt": "0.000005", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/flex", + "throughput_last_30m": null, + "uptime_last_1d": 99.97379717104731, + "uptime_last_30m": 99.92076767321068, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-6-astra", + "model_name": "OpenAI: GPT-6 Astra", + "name": "Azure | openai/gpt-6-astra-20260903", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.000001", + "input_cache_write": "0.0000125", + "overrides": [ + { + "completion": "0.000075", + "input_cache_read": "0.000002", + "input_cache_write": "0.000025", + "min_prompt_tokens": 272000, + "prompt": "0.00002" + } + ], + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure", + "throughput_last_30m": null, + "uptime_last_1d": 99.08610821237707, + "uptime_last_30m": 99.90611210215003, + "uptime_last_5m": 99.9481865284974 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-6-astra", + "model_name": "OpenAI: GPT-6 Astra", + "name": "OpenAI | openai/gpt-6-astra-20260903", + "pricing": { + "completion": "0.00005", + "discount": 0, + "input_cache_read": "0.000001", + "input_cache_write": "0.0000125", + "overrides": [ + { + "completion": "0.000075", + "input_cache_read": "0.000002", + "input_cache_write": "0.000025", + "min_prompt_tokens": 272000, + "prompt": "0.00002" + } + ], + "prompt": "0.00001", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai", + "throughput_last_30m": null, + "uptime_last_1d": 99.76041480148326, + "uptime_last_30m": 99.99254380514478, + "uptime_last_5m": 100 + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "openai/gpt-6-astra", + "model_name": "OpenAI: GPT-6 Astra", + "name": "Azure | openai/gpt-6-astra-20260903", + "pricing": { + "completion": "0.000055", + "discount": 0, + "input_cache_read": "0.0000011", + "input_cache_write": "0.00001375", + "overrides": [ + { + "completion": "0.0000825", + "input_cache_read": "0.0000022", + "input_cache_write": "0.0000275", + "min_prompt_tokens": 272000, + "prompt": "0.000022" + } + ], + "prompt": "0.000011", + "web_search": "0.01" + }, + "provider_name": "Azure", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_completion_tokens", + "reasoning", + "include_reasoning", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "azure/us", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 1050000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": 922000, + "model_id": "openai/gpt-6-astra", + "model_name": "OpenAI: GPT-6 Astra", + "name": "OpenAI | openai/gpt-6-astra-20260903", + "pricing": { + "completion": "0.0001", + "discount": 0, + "input_cache_read": "0.000002", + "input_cache_write": "0.000025", + "overrides": [ + { + "completion": "0.00015", + "input_cache_read": "0.000004", + "input_cache_write": "0.00005", + "min_prompt_tokens": 272000, + "prompt": "0.00004" + } + ], + "prompt": "0.00002", + "web_search": "0.01" + }, + "provider_name": "OpenAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "openai/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.99799347880612, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-6-astra", + "name": "OpenAI: GPT-6 Astra" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "GPT" + }, + "created": 1754414231, + "description": "gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "AkashML | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000017", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000003" + }, + "provider_name": "AkashML", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "stop", + "seed", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "akashml/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.48373373429366, + "uptime_last_30m": 99.9247554552295, + "uptime_last_5m": 99.93122420907841 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "CoreWeave | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000017", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000003" + }, + "provider_name": "CoreWeave", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.91980369005121, + "uptime_last_30m": 99.87767190316765, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "DekaLLM | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000018", + "discount": 0, + "prompt": "0.00000003" + }, + "provider_name": "DekaLLM", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "frequency_penalty", + "logit_bias", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "dekallm/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.81012854074996, + "uptime_last_30m": 99.75431861804222, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "DeepInfra | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000017", + "discount": 0, + "prompt": "0.000000037" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 98.39739088029438, + "uptime_last_30m": 99.99732691793638, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Crusoe | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000025", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000005" + }, + "provider_name": "Crusoe", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "min_p", + "repetition_penalty", + "top_k", + "logit_bias", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "crusoe/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.0561029753705, + "uptime_last_30m": 98.7048398091343, + "uptime_last_5m": 98.79253567508232 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Novita | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000025", + "discount": 0, + "prompt": "0.00000005" + }, + "provider_name": "Novita", + "quantization": "fp4", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 87.58191447978695, + "uptime_last_30m": 50.73529411764706, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Mancer 2 | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000003", + "discount": 0, + "prompt": "0.00000005" + }, + "provider_name": "Mancer 2", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "logit_bias", + "top_k", + "min_p", + "seed", + "top_a", + "response_format", + "logprobs", + "top_logprobs", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "mancer/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.28407165656684, + "uptime_last_30m": 99.95415998166399, + "uptime_last_5m": 99.83844911147011 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 4096, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "DigitalOcean | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000042", + "discount": 0, + "input_cache_read": "0.000000012", + "prompt": "0.00000006" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 99.98347697102837, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Google | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000036", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "repetition_penalty", + "top_k", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 94.62798505970817, + "uptime_last_30m": 95.05357503874924, + "uptime_last_5m": 94.92798998121478 + }, + { + "context_length": 128072, + "latency_last_30m": null, + "max_completion_tokens": 115264, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "BaseTen | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.0000001" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.91751736126508, + "uptime_last_30m": 99.75278121137207, + "uptime_last_5m": 100 + }, + { + "context_length": 128072, + "latency_last_30m": null, + "max_completion_tokens": 115264, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "BaseTen | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.0000001" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.90565881593538, + "uptime_last_30m": 99.63570127504553, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Parasail | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "input_cache_read": "0.000000055", + "prompt": "0.0000001" + }, + "provider_name": "Parasail", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.99129903056699, + "uptime_last_30m": 99.94982438534872, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "SambaNova | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000095", + "discount": 0, + "prompt": "0.00000014" + }, + "provider_name": "SambaNova", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sambanova", + "throughput_last_30m": null, + "uptime_last_1d": 97.32549263520963, + "uptime_last_30m": 99.528672427337, + "uptime_last_5m": 95 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Amazon Bedrock | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/eu-west-1", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Nebius | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Nebius", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "repetition_penalty", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "nebius/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.43097916796937, + "uptime_last_30m": 98.93617021276596, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Amazon Bedrock | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock", + "throughput_last_30m": null, + "uptime_last_1d": 98.23124158633945, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "DeepInfra | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/turbo", + "throughput_last_30m": null, + "uptime_last_1d": 99.98488533465243, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "SiliconFlow | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000015" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 91.8222360506119, + "uptime_last_30m": 86.2796833773087, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Phala | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 89.10528627910382, + "uptime_last_30m": 67.00507614213198, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Together | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 87.68450556541377, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Groq | openai/gpt-oss-120b", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000075", + "prompt": "0.00000015" + }, + "provider_name": "Groq", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "seed", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "groq", + "throughput_last_30m": null, + "uptime_last_1d": 99.97847608836703, + "uptime_last_30m": 99.9965962081759, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Mara | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Mara", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "top_k", + "top_p", + "tool_choice", + "tools", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "mara", + "throughput_last_30m": null, + "uptime_last_1d": 90.63242710567503, + "uptime_last_30m": 90.73359073359073, + "uptime_last_5m": 90.54054054054053 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "DeepInfra | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000095", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 63.92830056049872, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 40960, + "max_prompt_tokens": null, + "model_id": "openai/gpt-oss-120b", + "model_name": "OpenAI: gpt-oss-120b", + "name": "Cerebras | openai/gpt-oss-120b", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "input_cache_read": "0.00000035", + "prompt": "0.00000035" + }, + "provider_name": "Cerebras", + "quantization": "fp16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "frequency_penalty", + "presence_penalty", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cerebras/fp16", + "throughput_last_30m": null, + "uptime_last_1d": 99.99328511927988, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "openai/gpt-oss-120b", + "name": "OpenAI: gpt-oss-120b" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "chatml", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1726704000, + "description": "Qwen2.5 72B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and...", + "endpoints": [ + { + "context_length": 32768, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen-2.5-72b-instruct", + "model_name": "Qwen2.5 72B Instruct", + "name": "DeepInfra | qwen/qwen-2.5-72b-instruct", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "prompt": "0.00000036" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.88546998343605, + "uptime_last_30m": 99.96275882615821, + "uptime_last_5m": 100 + }, + { + "context_length": 32000, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": null, + "model_id": "qwen/qwen-2.5-72b-instruct", + "model_name": "Qwen2.5 72B Instruct", + "name": "Novita | qwen/qwen-2.5-72b-instruct", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "prompt": "0.00000038" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 92.95659295659296, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen-2.5-72b-instruct", + "name": "Qwen2.5 72B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "chatml", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1729036800, + "description": "Qwen2.5 7B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and...", + "endpoints": [ + { + "context_length": 32768, + "latency_last_30m": null, + "max_completion_tokens": 29491, + "max_prompt_tokens": null, + "model_id": "qwen/qwen-2.5-7b-instruct", + "model_name": "Qwen: Qwen2.5 7B Instruct", + "name": "Phala | qwen/qwen-2.5-7b-instruct", + "pricing": { + "completion": "0.0000002", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.99991756615728, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen-2.5-7b-instruct", + "name": "Qwen: Qwen2.5 7B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1738409840, + "description": "Qwen-Plus, based on the Qwen2.5 foundation model, is a 131K context model with a balanced performance, speed, and cost combination.", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 995904, + "model_id": "qwen/qwen-plus", + "model_name": "Qwen: Qwen-Plus", + "name": "Alibaba | qwen/qwen-plus-2025-01-25", + "pricing": { + "completion": "0.00000078", + "discount": 0, + "input_cache_read": "0.000000052", + "input_cache_write": "0.000000325", + "overrides": [ + { + "completion": "0.00000234", + "input_cache_read": "0.000000156", + "input_cache_write": "0.000000975", + "min_prompt_tokens": 256000, + "prompt": "0.00000078" + } + ], + "prompt": "0.00000026" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen-plus", + "name": "Qwen: Qwen-Plus" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1757347599, + "description": "Qwen Plus 0728, based on the Qwen3 foundation model, is a 1 million context hybrid reasoning model with a balanced performance, speed, and cost combination.", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 995904, + "model_id": "qwen/qwen-plus-2025-07-28", + "model_name": "Qwen: Qwen Plus 0728", + "name": "Alibaba | qwen/qwen-plus-2025-07-28", + "pricing": { + "completion": "0.00000078", + "discount": 0, + "overrides": [ + { + "completion": "0.00000234", + "min_prompt_tokens": 256000, + "prompt": "0.00000078" + } + ], + "prompt": "0.00000026" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen-plus-2025-07-28", + "name": "Qwen: Qwen Plus 0728" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1738410311, + "description": "Qwen2.5-VL is proficient in recognizing common objects such as flowers, birds, fish, and insects. It is also highly capable of analyzing texts, charts, icons, graphics, and layouts within images.", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 115200, + "max_prompt_tokens": null, + "model_id": "qwen/qwen2.5-vl-72b-instruct", + "model_name": "Qwen: Qwen2.5 VL 72B Instruct", + "name": "Parasail | qwen/qwen2.5-vl-72b-instruct", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.0000004", + "prompt": "0.0000008" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.99950714394846, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen2.5-vl-72b-instruct", + "name": "Qwen: Qwen2.5 VL 72B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "qwen3", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1745876478, + "description": "Qwen3-14B is a dense 14.8B parameter causal language model from the Qwen3 series, designed for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...", + "endpoints": [ + { + "context_length": 40960, + "latency_last_30m": null, + "max_completion_tokens": 36864, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-14b", + "model_name": "Qwen: Qwen3 14B", + "name": "NextBit | qwen/qwen3-14b-04-28", + "pricing": { + "completion": "0.00000022", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "NextBit", + "quantization": "int4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nextbit/int4", + "throughput_last_30m": null, + "uptime_last_1d": 99.06081977515537, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 40960, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-14b", + "model_name": "Qwen: Qwen3 14B", + "name": "DeepInfra | qwen/qwen3-14b-04-28", + "pricing": { + "completion": "0.00000024", + "discount": 0, + "prompt": "0.00000012" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.04853157449084, + "uptime_last_30m": 99.76744186046511, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": 98304, + "model_id": "qwen/qwen3-14b", + "model_name": "Qwen: Qwen3 14B", + "name": "Alibaba | qwen/qwen3-14b-04-28", + "pricing": { + "completion": "0.00000091", + "discount": 0, + "prompt": "0.0000002275" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-14b", + "name": "Qwen: Qwen3 14B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "qwen3", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1745875757, + "description": "Qwen3-235B-A22B is a 235B parameter mixture-of-experts (MoE) model developed by Qwen, activating 22B parameters per forward pass. It supports seamless switching between a \"thinking\" mode for complex reasoning, math, and...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": 98304, + "model_id": "qwen/qwen3-235b-a22b", + "model_name": "Qwen: Qwen3 235B A22B", + "name": "Alibaba | qwen/qwen3-235b-a22b-04-28", + "pricing": { + "completion": "0.00000182", + "discount": 0, + "prompt": "0.000000455" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-235b-a22b", + "name": "Qwen: Qwen3 235B A22B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1753119555, + "description": "Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following,...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "GMICloud | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000035", + "discount": 0.75, + "input_cache_read": "0.0000000175", + "prompt": "0.0000000875" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.86124621223658, + "uptime_last_30m": 98.73775236235574, + "uptime_last_5m": 99.38257489137892 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "DeepInfra | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000055", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.32179748140348, + "uptime_last_30m": 97.95876075018133, + "uptime_last_5m": 99.96832435856827 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Novita | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000058", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.97792084832419, + "uptime_last_30m": 99.60317460317461, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Parasail | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.0000008", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000014" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.34351801881701, + "uptime_last_30m": 99.91694926984567, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Alibaba | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.000000598", + "discount": 0, + "prompt": "0.0000001495" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99698183365679, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Venice | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.35469297916222, + "uptime_last_30m": 98.66468842729971, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Nebius | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Nebius", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "logit_bias", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nebius/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.60040350353218, + "uptime_last_30m": 98.21740214410372, + "uptime_last_5m": 97.55011135857461 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 32000, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "StreamLake | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000084", + "discount": 0.4, + "prompt": "0.00000021" + }, + "provider_name": "StreamLake", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop", + "presence_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake", + "throughput_last_30m": null, + "uptime_last_1d": 98.14435967171292, + "uptime_last_30m": 99.12010558732952, + "uptime_last_5m": 99.57264957264957 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Google | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.00000088", + "discount": 0, + "prompt": "0.00000022" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us-south1", + "throughput_last_30m": null, + "uptime_last_1d": 99.95469979261648, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-235b-a22b-2507", + "model_name": "Qwen: Qwen3 235B A22B Instruct 2507", + "name": "Google | qwen/qwen3-235b-a22b-07-25", + "pricing": { + "completion": "0.000001", + "discount": 0, + "prompt": "0.00000025" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us-south1", + "throughput_last_30m": null, + "uptime_last_1d": 99.93336313030758, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-235b-a22b-2507", + "name": "Qwen: Qwen3 235B A22B Instruct 2507" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "qwen3", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1745878604, + "description": "Qwen3, the latest generation in the Qwen large language model series, features both dense and mixture-of-experts (MoE) architectures to excel in reasoning, multilingual support, and advanced agent tasks. Its unique...", + "endpoints": [ + { + "context_length": 40960, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-30b-a3b", + "model_name": "Qwen: Qwen3 30B A3B", + "name": "DeepInfra | qwen/qwen3-30b-a3b-04-28", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "prompt": "0.00000012" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96587197113745, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": 98304, + "model_id": "qwen/qwen3-30b-a3b", + "model_name": "Qwen: Qwen3 30B A3B", + "name": "Alibaba | qwen/qwen3-30b-a3b-04-28", + "pricing": { + "completion": "0.00000052", + "discount": 0, + "prompt": "0.00000013" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.98788539584469, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-30b-a3b", + "name": "Qwen: Qwen3 30B A3B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1753806965, + "description": "Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language model from Qwen, with 3.3B active parameters per inference. It operates in non-thinking mode and is designed for high-quality instruction following, multilingual understanding, and...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 32000, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-30b-a3b-instruct-2507", + "model_name": "Qwen: Qwen3 30B A3B Instruct 2507", + "name": "StreamLake | qwen/qwen3-30b-a3b-instruct-2507", + "pricing": { + "completion": "0.00000019305", + "discount": 0.55, + "prompt": "0.00000004815" + }, + "provider_name": "StreamLake", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop", + "presence_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake", + "throughput_last_30m": null, + "uptime_last_1d": 98.15340336382047, + "uptime_last_30m": 98.53856171321083, + "uptime_last_5m": 98.92118669463589 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-30b-a3b-instruct-2507", + "model_name": "Qwen: Qwen3 30B A3B Instruct 2507", + "name": "DekaLLM | qwen/qwen3-30b-a3b-instruct-2507", + "pricing": { + "completion": "0.0000003", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "DekaLLM", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "frequency_penalty", + "logit_bias", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "dekallm", + "throughput_last_30m": null, + "uptime_last_1d": 99.32963744751919, + "uptime_last_30m": 99.88518943742825, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-30b-a3b-instruct-2507", + "model_name": "Qwen: Qwen3 30B A3B Instruct 2507", + "name": "SiliconFlow | qwen/qwen3-30b-a3b-instruct-2507", + "pricing": { + "completion": "0.0000003", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.08830205370758, + "uptime_last_30m": 98.47161572052401, + "uptime_last_5m": 99.92169146436962 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-30b-a3b-instruct-2507", + "model_name": "Qwen: Qwen3 30B A3B Instruct 2507", + "name": "Nebius | qwen/qwen3-30b-a3b-instruct-2507", + "pricing": { + "completion": "0.0000003", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "Nebius", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "repetition_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nebius/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.17118024257896, + "uptime_last_30m": 93.23781711991239, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-30b-a3b-instruct-2507", + "model_name": "Qwen: Qwen3 30B A3B Instruct 2507", + "name": "Alibaba | qwen/qwen3-30b-a3b-instruct-2507", + "pricing": { + "completion": "0.00000052", + "discount": 0, + "prompt": "0.00000013" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.9594175687701, + "uptime_last_30m": 99.94519046314059, + "uptime_last_5m": 99.87562189054727 + } + ], + "id": "qwen/qwen3-30b-a3b-instruct-2507", + "name": "Qwen: Qwen3 30B A3B Instruct 2507" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "qwen3", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1745875945, + "description": "Qwen3-32B is a dense 32.8B parameter causal language model from the Qwen3 series, optimized for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...", + "endpoints": [ + { + "context_length": 40960, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-32b", + "model_name": "Qwen: Qwen3 32B", + "name": "DeepInfra | qwen/qwen3-32b-04-28", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "prompt": "0.00000008" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.53182607159094, + "uptime_last_30m": 99.64178884006057, + "uptime_last_5m": 99.87834549878345 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-32b", + "model_name": "Qwen: Qwen3 32B", + "name": "SiliconFlow | qwen/qwen3-32b-04-28", + "pricing": { + "completion": "0.00000057", + "discount": 0, + "prompt": "0.00000014" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 85.99481420518403, + "uptime_last_30m": 97.00647249190939, + "uptime_last_5m": 95.08196721311475 + } + ], + "id": "qwen/qwen3-32b", + "name": "Qwen: Qwen3 32B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": "qwen3", + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1745876632, + "description": "Qwen3-8B is a dense 8.2B parameter causal language model from the Qwen3 series, designed for both reasoning-heavy tasks and efficient dialogue. It supports seamless switching between \"thinking\" mode for math,...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 8192, + "max_prompt_tokens": 98304, + "model_id": "qwen/qwen3-8b", + "model_name": "Qwen: Qwen3 8B", + "name": "Alibaba | qwen/qwen3-8b-04-28", + "pricing": { + "completion": "0.000000455", + "discount": 0, + "prompt": "0.000000117" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.81890904033884, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-8b", + "name": "Qwen: Qwen3 8B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1753230546, + "description": "Qwen3-Coder-480B-A35B-Instruct is a Mixture-of-Experts (MoE) code generation model developed by the Qwen team. It is optimized for agentic coding tasks such as function calling, tool use, and long-context reasoning over...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder", + "model_name": "Qwen: Qwen3 Coder 480B A35B", + "name": "Google | qwen/qwen3-coder-480b-a35b-07-25", + "pricing": { + "completion": "0.0000018", + "discount": 0, + "prompt": "0.00000022" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/us-south1", + "throughput_last_30m": null, + "uptime_last_1d": 99.84737782107484, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder", + "model_name": "Qwen: Qwen3 Coder 480B A35B", + "name": "DeepInfra | qwen/qwen3-coder-480b-a35b-07-25", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.0000003" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/turbo", + "throughput_last_30m": null, + "uptime_last_1d": 96.92155174662805, + "uptime_last_30m": 89.24050632911393, + "uptime_last_5m": null + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder", + "model_name": "Qwen: Qwen3 Coder 480B A35B", + "name": "Venice | qwen/qwen3-coder-480b-a35b-07-25", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.00000004", + "prompt": "0.00000035" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 93.21298385696927, + "uptime_last_30m": 96.49446494464945, + "uptime_last_5m": 90.9090909090909 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder", + "model_name": "Qwen: Qwen3 Coder 480B A35B", + "name": "Novita | qwen/qwen3-coder-480b-a35b-07-25", + "pricing": { + "completion": "0.00000155", + "discount": 0, + "prompt": "0.00000038" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 93.65969743328235, + "uptime_last_30m": 97.17607973421927, + "uptime_last_5m": 95.91836734693877 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 204800, + "model_id": "qwen/qwen3-coder", + "model_name": "Qwen: Qwen3 Coder 480B A35B", + "name": "Alibaba | qwen/qwen3-coder-480b-a35b-07-25", + "pricing": { + "completion": "0.000004875", + "discount": 0, + "overrides": [ + { + "completion": "0.000008775", + "min_prompt_tokens": 32000, + "prompt": "0.000001755" + }, + { + "completion": "0.000014625", + "min_prompt_tokens": 128000, + "prompt": "0.000002925" + } + ], + "prompt": "0.000000975" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba/opensource", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-coder", + "name": "Qwen: Qwen3 Coder 480B A35B" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1753972379, + "description": "Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...", + "endpoints": [ + { + "context_length": 160000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 Coder 30B A3B Instruct", + "name": "Novita | qwen/qwen3-coder-30b-a3b-instruct", + "pricing": { + "completion": "0.00000027", + "discount": 0, + "prompt": "0.00000007" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.16367940480217, + "uptime_last_30m": 99.83164983164983, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 Coder 30B A3B Instruct", + "name": "SiliconFlow | qwen/qwen3-coder-30b-a3b-instruct", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "prompt": "0.00000007" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.91178140350011, + "uptime_last_30m": 99.22106824925817, + "uptime_last_5m": 100 + }, + { + "context_length": 0, + "latency_last_30m": null, + "max_completion_tokens": 0, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 Coder 30B A3B Instruct", + "name": "Amazon Bedrock | qwen/qwen3-coder-30b-a3b-instruct", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "tool_choice", + "tools" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock", + "throughput_last_30m": null, + "uptime_last_1d": null, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 204800, + "model_id": "qwen/qwen3-coder-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 Coder 30B A3B Instruct", + "name": "Alibaba | qwen/qwen3-coder-30b-a3b-instruct", + "pricing": { + "completion": "0.0000014625", + "discount": 0, + "overrides": [ + { + "completion": "0.0000024375", + "min_prompt_tokens": 32000, + "prompt": "0.0000004875" + }, + { + "completion": "0.0000039", + "min_prompt_tokens": 128000, + "prompt": "0.00000078" + } + ], + "prompt": "0.0000002925" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-coder-30b-a3b-instruct", + "name": "Qwen: Qwen3 Coder 30B A3B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1770164101, + "description": "Qwen3-Coder-Next is an open-weight causal language model optimized for coding agents and local development workflows. It uses a sparse MoE design with 80B total parameters and only 3B activated per...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-next", + "model_name": "Qwen: Qwen3 Coder Next", + "name": "Parasail | qwen/qwen3-coder-next-2025-02-03", + "pricing": { + "completion": "0.0000008", + "discount": 0, + "input_cache_read": "0.00000007", + "prompt": "0.00000012" + }, + "provider_name": "Parasail", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.96916625555008, + "uptime_last_30m": 99.96581196581197, + "uptime_last_5m": 100 + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 64000, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-next", + "model_name": "Qwen: Qwen3 Coder Next", + "name": "StreamLake | qwen/qwen3-coder-next-2025-02-03", + "pricing": { + "completion": "0.0000009", + "discount": 0.4, + "input_cache_read": "0.000000036", + "prompt": "0.00000018" + }, + "provider_name": "StreamLake", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "tools", + "tool_choice", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "presence_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "streamlake", + "throughput_last_30m": null, + "uptime_last_1d": 98.91277691490231, + "uptime_last_30m": 98.30508474576271, + "uptime_last_5m": 96.96969696969697 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-coder-next", + "model_name": "Qwen: Qwen3 Coder Next", + "name": "Novita | qwen/qwen3-coder-next-2025-02-03", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.42568029536442, + "uptime_last_30m": 99.03381642512076, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 204800, + "model_id": "qwen/qwen3-coder-next", + "model_name": "Qwen: Qwen3 Coder Next", + "name": "Alibaba | qwen/qwen3-coder-next-2025-02-03", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "overrides": [ + { + "completion": "0.0000025", + "min_prompt_tokens": 32000, + "prompt": "0.0000005" + }, + { + "completion": "0.000004", + "min_prompt_tokens": 128000, + "prompt": "0.0000008" + } + ], + "prompt": "0.0000003" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.18157017277963, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-coder-next", + "name": "Qwen: Qwen3 Coder Next" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1758662707, + "description": "Qwen3 Coder Plus is Alibaba's proprietary version of the Open Source Qwen3 Coder 480B A35B. It is a powerful coding agent model specializing in autonomous programming via tool calling and...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 997952, + "model_id": "qwen/qwen3-coder-plus", + "model_name": "Qwen: Qwen3 Coder Plus", + "name": "Alibaba | qwen/qwen3-coder-plus", + "pricing": { + "completion": "0.00000325", + "discount": 0, + "input_cache_read": "0.00000013", + "input_cache_write": "0.0000008125", + "overrides": [ + { + "completion": "0.00000585", + "input_cache_read": "0.000000234", + "input_cache_write": "0.0000014625", + "min_prompt_tokens": 32000, + "prompt": "0.00000117" + }, + { + "completion": "0.00000975", + "input_cache_read": "0.00000039", + "input_cache_write": "0.0000024375", + "min_prompt_tokens": 128000, + "prompt": "0.00000195" + } + ], + "prompt": "0.00000065" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-coder-plus", + "name": "Qwen: Qwen3 Coder Plus" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1770671901, + "description": "Qwen3-Max-Thinking is the flagship reasoning model in the Qwen3 series, designed for high-stakes cognitive tasks that require deep, multi-step reasoning. By significantly scaling model capacity and reinforcement learning compute, it...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3-max-thinking", + "model_name": "Qwen: Qwen3 Max Thinking", + "name": "Alibaba | qwen/qwen3-max-thinking-20260123", + "pricing": { + "completion": "0.0000039", + "discount": 0, + "overrides": [ + { + "completion": "0.0000078", + "min_prompt_tokens": 32000, + "prompt": "0.00000156" + }, + { + "completion": "0.00000975", + "min_prompt_tokens": 128000, + "prompt": "0.00000195" + } + ], + "prompt": "0.00000078" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-max-thinking", + "name": "Qwen: Qwen3 Max Thinking" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1757612213, + "description": "Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-Next series optimized for fast, stable responses without \u201cthinking\u201d traces. It targets complex tasks across reasoning, code generation, knowledge QA, and multilingual...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-next-80b-a3b-instruct", + "model_name": "Qwen: Qwen3 Next 80B A3B Instruct", + "name": "DeepInfra | qwen/qwen3-next-80b-a3b-instruct-2509", + "pricing": { + "completion": "0.0000011", + "discount": 0, + "prompt": "0.00000009" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.85836360914942, + "uptime_last_30m": 91.61073825503355, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-next-80b-a3b-instruct", + "model_name": "Qwen: Qwen3 Next 80B A3B Instruct", + "name": "Alibaba | qwen/qwen3-next-80b-a3b-instruct-2509", + "pricing": { + "completion": "0.00000078", + "discount": 0, + "prompt": "0.0000000975" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.9976910644193, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-next-80b-a3b-instruct", + "model_name": "Qwen: Qwen3 Next 80B A3B Instruct", + "name": "Parasail | qwen/qwen3-next-80b-a3b-instruct-2509", + "pricing": { + "completion": "0.0000011", + "discount": 0, + "input_cache_read": "0.00000007", + "prompt": "0.0000001" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.979883323275, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-next-80b-a3b-instruct", + "model_name": "Qwen: Qwen3 Next 80B A3B Instruct", + "name": "Google | qwen/qwen3-next-80b-a3b-instruct-2509", + "pricing": { + "completion": "0.0000012", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Google", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "google-vertex/global", + "throughput_last_30m": null, + "uptime_last_1d": 99.70339949806069, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-next-80b-a3b-instruct", + "model_name": "Qwen: Qwen3 Next 80B A3B Instruct", + "name": "Novita | qwen/qwen3-next-80b-a3b-instruct-2509", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.91380796414411, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3-next-80b-a3b-instruct", + "name": "Qwen: Qwen3 Next 80B A3B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1758668687, + "description": "Qwen3-VL-235B-A22B Instruct is an open-weight multimodal model that unifies strong text generation with visual understanding across images and video. The Instruct model targets general vision-language use (VQA, document parsing, chart/table...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-235b-a22b-instruct", + "model_name": "Qwen: Qwen3 VL 235B A22B Instruct", + "name": "DeepInfra | qwen/qwen3-vl-235b-a22b-instruct", + "pricing": { + "completion": "0.00000088", + "discount": 0, + "input_cache_read": "0.00000011", + "prompt": "0.0000002" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.44459674868354, + "uptime_last_30m": 96.82539682539682, + "uptime_last_5m": 100 + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-235b-a22b-instruct", + "model_name": "Qwen: Qwen3 VL 235B A22B Instruct", + "name": "Venice | qwen/qwen3-vl-235b-a22b-instruct", + "pricing": { + "completion": "0.0000019", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.00000021" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.67336733673368, + "uptime_last_30m": 91.15789473684211, + "uptime_last_5m": 96.29629629629629 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-235b-a22b-instruct", + "model_name": "Qwen: Qwen3 VL 235B A22B Instruct", + "name": "Parasail | qwen/qwen3-vl-235b-a22b-instruct", + "pricing": { + "completion": "0.0000019", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.00000021" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.54640230482713, + "uptime_last_30m": 99.7165131112686, + "uptime_last_5m": 99.44707740916272 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-vl-235b-a22b-instruct", + "model_name": "Qwen: Qwen3 VL 235B A22B Instruct", + "name": "Alibaba | qwen/qwen3-vl-235b-a22b-instruct", + "pricing": { + "completion": "0.00000104", + "discount": 0, + "prompt": "0.00000026" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.68989745376074, + "uptime_last_30m": 99.62237432145386, + "uptime_last_5m": 99.55022488755623 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-235b-a22b-instruct", + "model_name": "Qwen: Qwen3 VL 235B A22B Instruct", + "name": "Novita | qwen/qwen3-vl-235b-a22b-instruct", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "prompt": "0.0000003" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": -2, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 94.49698937825588, + "uptime_last_30m": 92.71597967250142, + "uptime_last_5m": 93.33333333333333 + } + ], + "id": "qwen/qwen3-vl-235b-a22b-instruct", + "name": "Qwen: Qwen3 VL 235B A22B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1759794476, + "description": "Qwen3-VL-30B-A3B-Instruct is a multimodal model that unifies strong text generation with visual understanding for images and videos. Its Instruct variant optimizes instruction-following for general multimodal tasks. It excels in perception...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-vl-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 VL 30B A3B Instruct", + "name": "Alibaba | qwen/qwen3-vl-30b-a3b-instruct", + "pricing": { + "completion": "0.00000052", + "discount": 0, + "prompt": "0.00000013" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.93720633724566, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 VL 30B A3B Instruct", + "name": "DeepInfra | qwen/qwen3-vl-30b-a3b-instruct", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 93.25669917211282, + "uptime_last_30m": 88.88888888888889, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 VL 30B A3B Instruct", + "name": "Novita | qwen/qwen3-vl-30b-a3b-instruct", + "pricing": { + "completion": "0.0000007", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 97.94193052172798, + "uptime_last_30m": 98.75089587386096, + "uptime_last_5m": 99.1063829787234 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-30b-a3b-instruct", + "model_name": "Qwen: Qwen3 VL 30B A3B Instruct", + "name": "SiliconFlow | qwen/qwen3-vl-30b-a3b-instruct", + "pricing": { + "completion": "0.000001", + "discount": 0, + "prompt": "0.00000029" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "structured_outputs", + "response_format", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 90.56730928575374, + "uptime_last_30m": 94.09804854831033, + "uptime_last_5m": 91.3907284768212 + } + ], + "id": "qwen/qwen3-vl-30b-a3b-instruct", + "name": "Qwen: Qwen3 VL 30B A3B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1761231332, + "description": "Qwen3-VL-32B-Instruct is a large-scale multimodal vision-language model designed for high-precision understanding and reasoning across text, images, and video. With 32 billion parameters, it combines deep visual perception with advanced text...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-vl-32b-instruct", + "model_name": "Qwen: Qwen3 VL 32B Instruct", + "name": "Alibaba | qwen/qwen3-vl-32b-instruct", + "pricing": { + "completion": "0.000000416", + "discount": 0, + "prompt": "0.000000104" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99897915548937, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-vl-32b-instruct", + "name": "Qwen: Qwen3 VL 32B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "image", + "text" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1760463308, + "description": "Qwen3-VL-8B-Instruct is a multimodal vision-language model from the Qwen3-VL series, built for high-fidelity understanding and reasoning across text, images, and video. It features improved multimodal fusion with Interleaved-MRoPE for long-horizon...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": 129024, + "model_id": "qwen/qwen3-vl-8b-instruct", + "model_name": "Qwen: Qwen3 VL 8B Instruct", + "name": "Alibaba | qwen/qwen3-vl-8b-instruct", + "pricing": { + "completion": "0.000000455", + "discount": 0, + "prompt": "0.000000117" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.98857882008588, + "uptime_last_30m": 99.98886662213316, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3-vl-8b-instruct", + "model_name": "Qwen: Qwen3 VL 8B Instruct", + "name": "Parasail | qwen/qwen3-vl-8b-instruct", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "input_cache_read": "0.00000012", + "prompt": "0.00000025" + }, + "provider_name": "Parasail", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.11865935056795, + "uptime_last_30m": 99.80360065466448, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3-vl-8b-instruct", + "name": "Qwen: Qwen3 VL 8B Instruct" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1772053789, + "description": "The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-122b-a10b", + "model_name": "Qwen: Qwen3.5-122B-A10B", + "name": "SiliconFlow | qwen/qwen3.5-122b-a10b-20260224", + "pricing": { + "completion": "0.00000208", + "discount": 0, + "prompt": "0.00000026" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.20094628444197, + "uptime_last_30m": 63.93643031784841, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3.5-122b-a10b", + "model_name": "Qwen: Qwen3.5-122B-A10B", + "name": "Alibaba | qwen/qwen3.5-122b-a10b-20260224", + "pricing": { + "completion": "0.00000208", + "discount": 0, + "prompt": "0.00000026" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 97.29896951206503, + "uptime_last_30m": 98.26266195524146, + "uptime_last_5m": 90.6687402799378 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-122b-a10b", + "model_name": "Qwen: Qwen3.5-122B-A10B", + "name": "DeepInfra | qwen/qwen3.5-122b-a10b-20260224", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "prompt": "0.00000029" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.96738022426096, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-122b-a10b", + "model_name": "Qwen: Qwen3.5-122B-A10B", + "name": "AtlasCloud | qwen/qwen3.5-122b-a10b-20260224", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.0000003" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.53898352028804, + "uptime_last_30m": 99, + "uptime_last_5m": 98.98648648648648 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-122b-a10b", + "model_name": "Qwen: Qwen3.5-122B-A10B", + "name": "Novita | qwen/qwen3.5-122b-a10b-20260224", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "prompt": "0.0000004" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.91057455846189, + "uptime_last_30m": 99.9110056363097, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-122b-a10b", + "name": "Qwen: Qwen3.5-122B-A10B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1772053810, + "description": "The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance. Its overall capabilities are comparable to those of...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "Alibaba | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.00000156", + "discount": 0, + "prompt": "0.000000195" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "top_logprobs", + "logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 97.96765802751575, + "uptime_last_30m": 98.5153764581124, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "SiliconFlow | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.000002", + "discount": 0, + "prompt": "0.00000025" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.15222094947181, + "uptime_last_30m": 96.78111587982833, + "uptime_last_5m": 98.82352941176471 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "DeepInfra | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.0000026", + "discount": 0, + "prompt": "0.00000026" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.49212798374809, + "uptime_last_30m": 99.88370966178893, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "AtlasCloud | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.00000216", + "discount": 0, + "input_cache_read": "0.00000027", + "prompt": "0.00000027" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.63025821311675, + "uptime_last_30m": 99.26253687315634, + "uptime_last_5m": 98.13084112149532 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "Phala | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.0000003" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "logprobs", + "top_logprobs", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 97.69371196754564, + "uptime_last_30m": 98.28326180257511, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-27b", + "model_name": "Qwen: Qwen3.5-27B", + "name": "Novita | qwen/qwen3.5-27b-20260224", + "pricing": { + "completion": "0.0000024", + "discount": 0, + "prompt": "0.0000003" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.72078856079195, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-27b", + "name": "Qwen: Qwen3.5-27B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1772053822, + "description": "The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "Darkbloom | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.00000075", + "discount": 0, + "prompt": "0.00000008" + }, + "provider_name": "Darkbloom", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "response_format", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "darkbloom/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.88280195844096, + "uptime_last_30m": 98.69281045751634, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "DeepInfra | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000014" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96957958993288, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "Parasail | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000015" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.92006174541044, + "uptime_last_30m": 98.57142857142858, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "Alibaba | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.0000013", + "discount": 0, + "prompt": "0.0000001625" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "top_logprobs", + "logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.95631627020373, + "uptime_last_30m": 99.76076555023924, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "AtlasCloud | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.0000018", + "discount": 0, + "input_cache_read": "0.000000225", + "prompt": "0.000000225" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.86184210526315, + "uptime_last_30m": 99.32386747802569, + "uptime_last_5m": 99.375 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "SiliconFlow | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.0000018", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000024" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.14016783851214, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-35b-a3b", + "model_name": "Qwen: Qwen3.5-35B-A3B", + "name": "Venice | qwen/qwen3.5-35b-a3b-20260224", + "pricing": { + "completion": "0.00000125", + "discount": 0, + "input_cache_read": "0.00000015625", + "prompt": "0.0000003125" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tool_choice", + "logprobs", + "response_format", + "structured_outputs", + "tools", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 99.89800471728184, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-35b-a3b", + "name": "Qwen: Qwen3.5-35B-A3B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1771223018, + "description": "The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "Alibaba | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.00000234", + "discount": 0, + "prompt": "0.00000039" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.95283018867924, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "DeepInfra | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.000003", + "discount": 0, + "input_cache_read": "0.00000022", + "prompt": "0.00000045" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.84939008444985, + "uptime_last_30m": 97.4477958236659, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "Parasail | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000036", + "discount": 0, + "input_cache_read": "0.0000003", + "prompt": "0.0000005" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.89948194541097, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "DigitalOcean | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000011", + "prompt": "0.00000055" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 97.18554289594809, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "Phala | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.000000225", + "prompt": "0.00000055" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.36381914729083, + "uptime_last_30m": 99.89071038251366, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "AtlasCloud | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.00000055", + "prompt": "0.00000055" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 94.87344560024395, + "uptime_last_30m": 99.34426229508196, + "uptime_last_5m": 100 + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 64000, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "StreamLake | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000036", + "discount": 0, + "input_cache_read": "0.00000012", + "prompt": "0.0000006" + }, + "provider_name": "StreamLake", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop", + "presence_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake", + "throughput_last_30m": null, + "uptime_last_1d": 95.35541195476576, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "GMICloud | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000036", + "discount": 0, + "prompt": "0.0000006" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "response_format", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.60177134963841, + "uptime_last_30m": 80.78078078078079, + "uptime_last_5m": 95.23809523809523 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "Novita | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000036", + "discount": 0, + "prompt": "0.0000006" + }, + "provider_name": "Novita", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita", + "throughput_last_30m": null, + "uptime_last_1d": 98.98934215362, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-397b-a17b", + "model_name": "Qwen: Qwen3.5 397B A17B", + "name": "Venice | qwen/qwen3.5-397b-a17b-20260216", + "pricing": { + "completion": "0.0000045", + "discount": 0, + "prompt": "0.00000075" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 97.11322976412049, + "uptime_last_30m": 96.4349376114082, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3.5-397b-a17b", + "name": "Qwen: Qwen3.5 397B A17B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1773152396, + "description": "Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "Darkbloom | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000013", + "discount": 0, + "prompt": "0.00000008" + }, + "provider_name": "Darkbloom", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "darkbloom/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.87083753784056, + "uptime_last_30m": 99.39393939393939, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "SiliconFlow | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000015", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.9808243133826, + "uptime_last_30m": 95.43730169678575, + "uptime_last_5m": 92.20945083014048 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "DeepInfra | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000015", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.84182112031911, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "Venice | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000015", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.83970791219555, + "uptime_last_30m": 99.98249146458899, + "uptime_last_5m": 99.82222222222222 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "Parasail | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000025", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "Parasail", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.92037371270835, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.5-9b", + "model_name": "Qwen: Qwen3.5-9B", + "name": "Together | qwen/qwen3.5-9b-20260310", + "pricing": { + "completion": "0.00000025", + "discount": 0, + "prompt": "0.00000017" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.49856866681537, + "uptime_last_30m": 99.69933854479855, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-9b", + "name": "Qwen: Qwen3.5-9B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1771229416, + "description": "The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher inference efficiency. In a variety of...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.5-plus-02-15", + "model_name": "Qwen: Qwen3.5 Plus 2026-02-15", + "name": "Alibaba | qwen/qwen3.5-plus-20260216", + "pricing": { + "completion": "0.00000156", + "discount": 0, + "overrides": [ + { + "completion": "0.00000195", + "min_prompt_tokens": 256000, + "prompt": "0.000000325" + } + ], + "prompt": "0.00000026" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99704124504409, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-plus-02-15", + "name": "Qwen: Qwen3.5 Plus 2026-02-15" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1777261368, + "description": "Qwen3.5 Plus (April 2026) is a large-scale multimodal language model from Alibaba. It accepts text, image, and video input and produces text output, with a 1M token context window. This...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.5-plus-20260420", + "model_name": "Qwen: Qwen3.5 Plus 2026-04-20", + "name": "Alibaba | qwen/qwen3.5-plus-20260420", + "pricing": { + "completion": "0.0000018", + "discount": 0, + "input_cache_write": "0.000000375", + "overrides": [ + { + "completion": "0.00000225", + "input_cache_write": "0.00000046875", + "min_prompt_tokens": 256000, + "prompt": "0.000000375" + } + ], + "prompt": "0.0000003" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.5-plus-20260420", + "name": "Qwen: Qwen3.5 Plus 2026-04-20" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1777255064, + "description": "Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026. It features hybrid multimodal capabilities \u2014 accepting text, image, and video inputs...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "Chutes | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.000002", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000003" + }, + "provider_name": "Chutes", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "chutes/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.92827899176244, + "uptime_last_30m": 99.03225806451613, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "SiliconFlow | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "prompt": "0.0000003" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.52319664771026, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 262140, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "Phala | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.0000027", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000032" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 98.94425080087845, + "uptime_last_30m": 98.81847475832438, + "uptime_last_5m": 98.91304347826086 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 81920, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "DeepInfra | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "prompt": "0.00000032" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "tools", + "tool_choice", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.59041291865091, + "uptime_last_30m": 99.8936170212766, + "uptime_last_5m": 100 + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "Venice | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.00000325", + "discount": 0, + "prompt": "0.000000325" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tool_choice", + "tools", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.8642725745821, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 258048, + "model_id": "qwen/qwen3.6-27b", + "model_name": "Qwen: Qwen3.6 27B", + "name": "Alibaba | qwen/qwen3.6-27b-20260422", + "pricing": { + "completion": "0.0000027", + "discount": 0, + "prompt": "0.00000045" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "top_logprobs", + "logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3.6-27b", + "name": "Qwen: Qwen3.6 27B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1777260255, + "description": "Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per token. It uses a hybrid sparse mixture-of-experts architecture combining Gated...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "Darkbloom | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.0000007", + "discount": 0, + "prompt": "0.00000005" + }, + "provider_name": "Darkbloom", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "darkbloom/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.93925343713971, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "AkashML | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.0000009", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.0000001" + }, + "provider_name": "AkashML", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "akashml/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.82929613455235, + "uptime_last_30m": 99.57284961966062, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "DeepInfra | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.00000095", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "logit_bias" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 94.95088905757598, + "uptime_last_30m": 99.38725490196079, + "uptime_last_5m": null + }, + { + "context_length": 256000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "Venice | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.000001", + "discount": 0, + "prompt": "0.0000001" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.89806983300802, + "uptime_last_30m": 99.91235758106923, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "Parasail | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.000001", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000015" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96063957242146, + "uptime_last_30m": 99.98235631423404, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "AtlasCloud | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.00000111375", + "discount": 0, + "input_cache_read": "0.000000186", + "prompt": "0.000000186" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.38732862615964, + "uptime_last_30m": 99.64912280701755, + "uptime_last_5m": 97.61904761904762 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "Io Net | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.00000099", + "discount": 0, + "input_cache_read": "0.000000099", + "prompt": "0.00000019" + }, + "provider_name": "Io Net", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "seed", + "stop", + "min_p", + "top_k", + "top_p", + "presence_penalty", + "frequency_penalty", + "repetition_penalty" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "io-net/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 83.00414937759336, + "uptime_last_30m": 83.03571428571429, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "Phala | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.00000127", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.66321431158109, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "SiliconFlow | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.0000018", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.00000024" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "repetition_penalty", + "presence_penalty", + "max_tokens", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.30310848998649, + "uptime_last_30m": 98.50299401197606, + "uptime_last_5m": 98.57142857142858 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.6-35b-a3b", + "model_name": "Qwen: Qwen3.6 35B A3B", + "name": "CoreWeave | qwen/qwen3.6-35b-a3b-20260415", + "pricing": { + "completion": "0.00000125", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.00000025" + }, + "provider_name": "CoreWeave", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.93826752210244, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.6-35b-a3b", + "name": "Qwen: Qwen3.6 35B A3B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1777261362, + "description": "Qwen3.6 Flash is a fast, efficient language model from Alibaba's Qwen 3.6 series. It supports text, image, and video input with a 1M token context window. Tiered pricing kicks in...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.6-flash", + "model_name": "Qwen: Qwen3.6 Flash", + "name": "Alibaba | qwen/qwen3.6-flash", + "pricing": { + "completion": "0.000001125", + "discount": 0, + "input_cache_write": "0.000000234375", + "overrides": [ + { + "completion": "0.000003", + "input_cache_write": "0.0000009375", + "min_prompt_tokens": 256000, + "prompt": "0.00000075" + } + ], + "prompt": "0.0000001875" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.6-flash", + "name": "Qwen: Qwen3.6 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1777260242, + "description": "Qwen3.6-Max-Preview is a proprietary frontier model from Alibaba Cloud built on a sparse mixture-of-experts architecture with approximately 1 trillion total parameters. It is optimized for agentic coding, tool use, and...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 229376, + "model_id": "qwen/qwen3.6-max-preview", + "model_name": "Qwen: Qwen3.6 Max Preview", + "name": "Alibaba | qwen/qwen3.6-max-preview-20260420", + "pricing": { + "completion": "0.000006162", + "discount": 0, + "input_cache_write": "0.00000128375", + "overrides": [ + { + "completion": "0.00000948", + "input_cache_write": "0.000001975", + "min_prompt_tokens": 128000, + "prompt": "0.00000158" + } + ], + "prompt": "0.000001027" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "top_logprobs", + "logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": null + } + ], + "id": "qwen/qwen3.6-max-preview", + "name": "Qwen: Qwen3.6 Max Preview" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen3" + }, + "created": 1775133557, + "description": "Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear attention with sparse mixture-of-experts routing, enabling strong scalability and high-performance inference. Compared to the 3.5 series, it delivers...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.6-plus", + "model_name": "Qwen: Qwen3.6 Plus", + "name": "Alibaba | qwen/qwen3.6-plus-04-02", + "pricing": { + "completion": "0.00000195", + "discount": 0, + "input_cache_write": "0.00000040625", + "overrides": [ + { + "completion": "0.0000039", + "input_cache_write": "0.000001625", + "min_prompt_tokens": 256000, + "prompt": "0.0000013" + } + ], + "prompt": "0.000000325" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tool_choice", + "tools", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.6-plus", + "name": "Qwen: Qwen3.6 Plus" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1785190561, + "description": "Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.7-flash", + "model_name": "Qwen: Qwen3.7 Flash", + "name": "Alibaba | qwen/qwen3.7-flash-20260727", + "pricing": { + "completion": "0.00000013", + "discount": 0, + "input_cache_read": "0.000000006", + "input_cache_write": "0.000000038", + "overrides": [ + { + "completion": "0.0000004", + "input_cache_read": "0.00000002", + "input_cache_write": "0.000000125", + "min_prompt_tokens": 32000, + "prompt": "0.0000001" + }, + { + "completion": "0.0000008", + "input_cache_read": "0.00000004", + "input_cache_write": "0.00000025", + "min_prompt_tokens": 256000, + "prompt": "0.0000002" + } + ], + "prompt": "0.00000003" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99972687183099, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.7-flash", + "name": "Qwen: Qwen3.7 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1779376861, + "description": "Qwen3.7-Max is the flagship model in Alibaba's Qwen3.7 series. It supports text input and output and is designed for agent-centric workloads, with particular strengths in coding, office and productivity tasks,...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.7-max", + "model_name": "Qwen: Qwen3.7 Max", + "name": "Alibaba | qwen/qwen3.7-max-20260520", + "pricing": { + "completion": "0.000004425", + "discount": 0, + "input_cache_read": "0.000000295", + "input_cache_write": "0.00000184375", + "prompt": "0.000001475" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.7-max", + "name": "Qwen: Qwen3.7 Max" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image" + ], + "instruct_type": null, + "modality": "text+image->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1780491783, + "description": "Qwen3.7-Plus is a cost-effective model in Alibaba's Qwen3.7 series. It supports text and image input with text output, building on the series' text capabilities with a comprehensive upgrade to its...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.7-plus", + "model_name": "Qwen: Qwen3.7 Plus", + "name": "Alibaba | qwen/qwen3.7-plus-20260602", + "pricing": { + "completion": "0.00000128", + "discount": 0, + "input_cache_read": "0.000000064", + "input_cache_write": "0.0000004", + "overrides": [ + { + "completion": "0.00000384", + "input_cache_read": "0.000000192", + "input_cache_write": "0.0000012", + "min_prompt_tokens": 256000, + "prompt": "0.00000096" + } + ], + "prompt": "0.00000032" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "structured_outputs", + "top_k", + "frequency_penalty", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99282283816395, + "uptime_last_30m": 99.9988178411415, + "uptime_last_5m": 99.99105065330231 + } + ], + "id": "qwen/qwen3.7-plus", + "name": "Qwen: Qwen3.7 Plus" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1786551702, + "description": "Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of [Qwen3.8 Max](/qwen/qwen3.8-max), with 95 billion active parameters out of 2.4 trillion total. It is...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "Novita | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.000002" + }, + "provider_name": "Novita", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita", + "throughput_last_30m": null, + "uptime_last_1d": 99.90973201692525, + "uptime_last_30m": 99.875, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "Alibaba | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.00000025", + "input_cache_write": "0.0000025", + "prompt": "0.000002" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "top_k", + "frequency_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.94134037248863, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "SiliconFlow | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.000002" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.92196410348761, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "Venice | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0.2, + "input_cache_read": "0.00000025", + "prompt": "0.000002" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 97.87638406874896, + "uptime_last_30m": 92.3469387755102, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "Modal | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.000002" + }, + "provider_name": "Modal", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "modal", + "throughput_last_30m": null, + "uptime_last_1d": 99.82009410462219, + "uptime_last_30m": 99.53298307063632, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "DeepInfra | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000002", + "prompt": "0.000002" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.86275733000623, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1010000, + "latency_last_30m": null, + "max_completion_tokens": 909000, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-2.4t-a95b", + "model_name": "Qwen: Qwen3.8 2.4T A95B", + "name": "Together | qwen/qwen3.8-2.4t-a95b-20260812", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.00000025", + "prompt": "0.000002" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.8-2.4t-a95b", + "name": "Qwen: Qwen3.8 2.4T A95B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1786722910, + "description": "Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "DeepInfra | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.000001875", + "discount": 0.25, + "input_cache_read": "0.0000000375", + "prompt": "0.00000015" + }, + "provider_name": "DeepInfra", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 90.83829399281375, + "uptime_last_30m": 97.09386806161, + "uptime_last_5m": 98.18688981868898 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Darkbloom | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.000002", + "discount": 0, + "prompt": "0.00000015" + }, + "provider_name": "Darkbloom", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "darkbloom/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 97.0504627353752, + "uptime_last_30m": 99.6779388083736, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Phala | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.000002075", + "discount": 0.17, + "input_cache_read": "0.0000000415", + "prompt": "0.0000001992" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 96.67581602373888, + "uptime_last_30m": 97.8076379066478, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "DekaLLM | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.0000002" + }, + "provider_name": "DekaLLM", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "frequency_penalty", + "logit_bias", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "dekallm", + "throughput_last_30m": null, + "uptime_last_1d": 99.34094042457686, + "uptime_last_30m": 98.49793834675044, + "uptime_last_5m": 97.53896636587368 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Mancer 2 | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "prompt": "0.0000002" + }, + "provider_name": "Mancer 2", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "logit_bias", + "top_k", + "min_p", + "seed", + "top_a", + "response_format", + "logprobs", + "top_logprobs", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "mancer/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.92895725985115, + "uptime_last_30m": 98.4573002754821, + "uptime_last_5m": 95.12820512820512 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Reka | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.00000255", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.000000214" + }, + "provider_name": "Reka", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "max_tokens", + "stop", + "seed", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "reka/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96801346801347, + "uptime_last_30m": 99.97826559443598, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Parasail | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000022", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000024" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.61001965669116, + "uptime_last_30m": 99.95998399359743, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Chutes | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000022", + "discount": 0, + "input_cache_read": "0.000000024", + "prompt": "0.00000024" + }, + "provider_name": "Chutes", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "chutes/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.75534460832908, + "uptime_last_30m": 99.34756820877817, + "uptime_last_5m": 98.66666666666667 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "AkashML | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000022", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000025" + }, + "provider_name": "AkashML", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "akashml/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.7024032623199, + "uptime_last_30m": 99.59893048128342, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Ionstream | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.00000255", + "discount": 0, + "input_cache_read": "0.0000001", + "prompt": "0.00000028" + }, + "provider_name": "Ionstream", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "ionstream/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.25742688706508, + "uptime_last_30m": 97.61904761904762, + "uptime_last_5m": 92.6829268292683 + }, + { + "context_length": 65536, + "latency_last_30m": null, + "max_completion_tokens": 58982, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Io Net | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000028", + "discount": 0, + "input_cache_read": "0.00000018", + "prompt": "0.0000003" + }, + "provider_name": "Io Net", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs", + "seed", + "stop", + "min_p", + "top_k", + "top_p", + "presence_penalty", + "frequency_penalty", + "repetition_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "io-net/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.89001039038148, + "uptime_last_30m": 98.84169884169884, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "CoreWeave | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.000003", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.0000004" + }, + "provider_name": "CoreWeave", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.87609817169253, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Novita | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.000003", + "discount": 0, + "input_cache_read": "0.000000085", + "prompt": "0.00000042" + }, + "provider_name": "Novita", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita", + "throughput_last_30m": null, + "uptime_last_1d": 99.40365520788006, + "uptime_last_30m": 99.81464318813717, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Alibaba | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.00000255", + "discount": 0, + "input_cache_read": "0.000000085", + "input_cache_write": "0.00000053125", + "prompt": "0.000000425" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "top_k", + "frequency_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.9545622754463, + "uptime_last_30m": 99.12559618441972, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Cloudflare | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000045" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "min_p", + "stop", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 91.39115778555686, + "uptime_last_30m": 82.73381294964028, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "qwen/qwen3.8-27b", + "model_name": "Qwen: Qwen3.8 27B", + "name": "Venice | qwen/qwen3.8-27b-20260814", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "prompt": "0.00000045" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.71960203093359, + "uptime_last_30m": 99.19928825622776, + "uptime_last_5m": 99.52606635071089 + } + ], + "id": "qwen/qwen3.8-27b", + "name": "Qwen: Qwen3.8 27B" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Qwen" + }, + "created": 1787773060, + "description": "Qwen3.8 Flash is a multimodal reasoning model from Alibaba. It is suited for coding assistance, agentic workflows, visual understanding, document and codebase analysis, desktop interaction, chart analysis, and long-video analysis.", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 983616, + "model_id": "qwen/qwen3.8-flash", + "model_name": "Qwen: Qwen3.8 Flash", + "name": "Alibaba | qwen/qwen3.8-flash-20260826", + "pricing": { + "completion": "0.00000047", + "discount": 0, + "input_cache_read": "0.000000016", + "input_cache_write": "0.0000002", + "prompt": "0.00000015" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "tools", + "structured_outputs", + "top_k", + "frequency_penalty", + "stop", + "logprobs", + "top_logprobs", + "tool_choice" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.99546361627671, + "uptime_last_30m": 99.99828728997893, + "uptime_last_5m": 100 + } + ], + "id": "qwen/qwen3.8-flash", + "name": "Qwen: Qwen3.8 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "audio" + ], + "instruct_type": null, + "modality": "text+image+audio->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1784325956, + "description": "Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...", + "endpoints": [ + { + "context_length": 524288, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "thinkingmachines/inkling", + "model_name": "Thinking Machines: Inkling", + "name": "DeepInfra | thinkingmachines/inkling-20260715", + "pricing": { + "completion": "0.00000405", + "discount": 0, + "input_cache_read": "0.00000016", + "prompt": "0.00000095" + }, + "provider_name": "DeepInfra", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.27203496457189, + "uptime_last_30m": 89.88095238095238, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "thinkingmachines/inkling", + "model_name": "Thinking Machines: Inkling", + "name": "BaseTen | thinkingmachines/inkling-20260715", + "pricing": { + "completion": "0.00000405", + "discount": 0, + "input_cache_read": "0.00000017", + "prompt": "0.000001" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 91.77924145283542, + "uptime_last_30m": 88.76889848812095, + "uptime_last_5m": 89.1891891891892 + }, + { + "context_length": 524288, + "latency_last_30m": null, + "max_completion_tokens": 471859, + "max_prompt_tokens": null, + "model_id": "thinkingmachines/inkling", + "model_name": "Thinking Machines: Inkling", + "name": "Together | thinkingmachines/inkling-20260715", + "pricing": { + "completion": "0.00000405", + "discount": 0, + "input_cache_read": "0.00000017", + "prompt": "0.000001" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 97.68535441901403, + "uptime_last_30m": 99.20382165605095, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 32768, + "max_prompt_tokens": null, + "model_id": "thinkingmachines/inkling", + "model_name": "Thinking Machines: Inkling", + "name": "BaseTen | thinkingmachines/inkling-20260715", + "pricing": { + "completion": "0.00000405", + "discount": 0, + "input_cache_read": "0.00000017", + "prompt": "0.000001" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 92.25062922241356, + "uptime_last_30m": 83.15789473684211, + "uptime_last_5m": 86.36363636363636 + } + ], + "id": "thinkingmachines/inkling", + "name": "Thinking Machines: Inkling" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Grok" + }, + "created": 1774979019, + "description": "Grok 4.20 is a reasoning model from SpaceXAI with industry-leading speed and agentic tool calling capabilities. It combines the lowest hallucination rate on the market with strict prompt adherance, delivering...", + "endpoints": [ + { + "context_length": 2000000, + "latency_last_30m": null, + "max_completion_tokens": 1800000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.20", + "model_name": "SpaceXAI: Grok 4.20", + "name": "xAI | x-ai/grok-4.20-20260309", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "input_cache_read": "0.0000002", + "overrides": [ + { + "completion": "0.000005", + "input_cache_read": "0.0000004", + "min_prompt_tokens": 200000, + "prompt": "0.0000025" + } + ], + "prompt": "0.00000125", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr", + "throughput_last_30m": null, + "uptime_last_1d": 99.82892405449991, + "uptime_last_30m": 99.81515711645102, + "uptime_last_5m": 100 + }, + { + "context_length": 2000000, + "latency_last_30m": null, + "max_completion_tokens": 1800000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.20", + "model_name": "SpaceXAI: Grok 4.20", + "name": "xAI | x-ai/grok-4.20-20260309", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "input_cache_read": "0.0000002", + "overrides": [ + { + "completion": "0.000005", + "input_cache_read": "0.0000004", + "min_prompt_tokens": 200000, + "prompt": "0.0000025" + } + ], + "prompt": "0.00000125", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai", + "throughput_last_30m": null, + "uptime_last_1d": 99.87455563273812, + "uptime_last_30m": 99.98896247240619, + "uptime_last_5m": 100 + }, + { + "context_length": 2000000, + "latency_last_30m": null, + "max_completion_tokens": 1800000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.20", + "model_name": "SpaceXAI: Grok 4.20", + "name": "xAI | x-ai/grok-4.20-20260309", + "pricing": { + "completion": "0.000005", + "discount": 0, + "input_cache_read": "0.0000004", + "overrides": [ + { + "completion": "0.00001", + "input_cache_read": "0.0000008", + "min_prompt_tokens": 200000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr/priority", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 2000000, + "latency_last_30m": null, + "max_completion_tokens": 1800000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.20", + "model_name": "SpaceXAI: Grok 4.20", + "name": "xAI | x-ai/grok-4.20-20260309", + "pricing": { + "completion": "0.000005", + "discount": 0, + "input_cache_read": "0.0000004", + "overrides": [ + { + "completion": "0.00001", + "input_cache_read": "0.0000008", + "min_prompt_tokens": 200000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/priority", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "x-ai/grok-4.20", + "name": "SpaceXAI: Grok 4.20" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Grok" + }, + "created": 1777591821, + "description": "Grok 4.3 is a reasoning model from SpaceXAI. It accepts text and image inputs with text output, and is suited for agentic workflows, instruction-following tasks, and applications requiring high factual...", + "endpoints": [ + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 900000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.3", + "model_name": "SpaceXAI: Grok 4.3", + "name": "xAI | x-ai/grok-4.3-20260430", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "input_cache_read": "0.0000002", + "overrides": [ + { + "completion": "0.000005", + "input_cache_read": "0.0000004", + "min_prompt_tokens": 200000, + "prompt": "0.0000025" + } + ], + "prompt": "0.00000125", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr", + "throughput_last_30m": null, + "uptime_last_1d": 99.72289048659604, + "uptime_last_30m": 99.94891443167305, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 900000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.3", + "model_name": "SpaceXAI: Grok 4.3", + "name": "xAI | x-ai/grok-4.3-20260430", + "pricing": { + "completion": "0.0000025", + "discount": 0, + "input_cache_read": "0.0000002", + "overrides": [ + { + "completion": "0.000005", + "input_cache_read": "0.0000004", + "min_prompt_tokens": 200000, + "prompt": "0.0000025" + } + ], + "prompt": "0.00000125", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai", + "throughput_last_30m": null, + "uptime_last_1d": 99.88792054374922, + "uptime_last_30m": 99.99608763693271, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 900000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.3", + "model_name": "SpaceXAI: Grok 4.3", + "name": "xAI | x-ai/grok-4.3-20260430", + "pricing": { + "completion": "0.000005", + "discount": 0, + "input_cache_read": "0.0000004", + "overrides": [ + { + "completion": "0.00001", + "input_cache_read": "0.0000008", + "min_prompt_tokens": 200000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr/priority", + "throughput_last_30m": null, + "uptime_last_1d": 98.63636363636363, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 900000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.3", + "model_name": "SpaceXAI: Grok 4.3", + "name": "xAI | x-ai/grok-4.3-20260430", + "pricing": { + "completion": "0.000005", + "discount": 0, + "input_cache_read": "0.0000004", + "overrides": [ + { + "completion": "0.00001", + "input_cache_read": "0.0000008", + "min_prompt_tokens": 200000, + "prompt": "0.000005" + } + ], + "prompt": "0.0000025", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/priority", + "throughput_last_30m": null, + "uptime_last_1d": 96.26749611197512, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "x-ai/grok-4.3", + "name": "SpaceXAI: Grok 4.3" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Grok" + }, + "created": 1783523154, + "description": "Grok 4.5 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM.", + "endpoints": [ + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.5", + "model_name": "SpaceXAI: Grok 4.5", + "name": "xAI | x-ai/grok-4.5-20260708", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000003", + "overrides": [ + { + "completion": "0.000012", + "input_cache_read": "0.0000006", + "min_prompt_tokens": 200000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr", + "throughput_last_30m": null, + "uptime_last_1d": 99.71886713869222, + "uptime_last_30m": 99.8689813298395, + "uptime_last_5m": 100 + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.5", + "model_name": "SpaceXAI: Grok 4.5", + "name": "xAI | x-ai/grok-4.5-20260708", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000003", + "overrides": [ + { + "completion": "0.000012", + "input_cache_read": "0.0000006", + "min_prompt_tokens": 200000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai", + "throughput_last_30m": null, + "uptime_last_1d": 98.8602851323829, + "uptime_last_30m": 99.3480032599837, + "uptime_last_5m": 99.64850615114236 + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.5", + "model_name": "SpaceXAI: Grok 4.5", + "name": "xAI | x-ai/grok-4.5-20260708", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.0000006", + "overrides": [ + { + "completion": "0.000024", + "input_cache_read": "0.0000012", + "min_prompt_tokens": 200000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr/priority", + "throughput_last_30m": null, + "uptime_last_1d": 96.53401797175867, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.5", + "model_name": "SpaceXAI: Grok 4.5", + "name": "xAI | x-ai/grok-4.5-20260708", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.0000006", + "overrides": [ + { + "completion": "0.000024", + "input_cache_read": "0.0000012", + "min_prompt_tokens": 200000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/priority", + "throughput_last_30m": null, + "uptime_last_1d": 94.8905109489051, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "x-ai/grok-4.5", + "name": "SpaceXAI: Grok 4.5" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "file" + ], + "instruct_type": null, + "modality": "text+image+file->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Grok" + }, + "created": 1786548957, + "description": "Grok 4.6 is SpaceXAI's smartest model with frontier performance on coding, knowledge work, and STEM.", + "endpoints": [ + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.6", + "model_name": "SpaceXAI: Grok 4.6", + "name": "xAI | x-ai/grok-4.6-20260810", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000005", + "overrides": [ + { + "completion": "0.000012", + "input_cache_read": "0.000001", + "min_prompt_tokens": 200000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr", + "throughput_last_30m": null, + "uptime_last_1d": 99.78355189868469, + "uptime_last_30m": 99.86468200270636, + "uptime_last_5m": 99.69135802469135 + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.6", + "model_name": "SpaceXAI: Grok 4.6", + "name": "xAI | x-ai/grok-4.6-20260810", + "pricing": { + "completion": "0.000006", + "discount": 0, + "input_cache_read": "0.0000005", + "overrides": [ + { + "completion": "0.000012", + "input_cache_read": "0.000001", + "min_prompt_tokens": 200000, + "prompt": "0.000004" + } + ], + "prompt": "0.000002", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai", + "throughput_last_30m": null, + "uptime_last_1d": 99.52041224455627, + "uptime_last_30m": 99.80021154072159, + "uptime_last_5m": 99.92663242846662 + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.6", + "model_name": "SpaceXAI: Grok 4.6", + "name": "Amazon Bedrock | x-ai/grok-4.6-20260810", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000055", + "input_cache_write": "0", + "overrides": [ + { + "completion": "0.0000132", + "input_cache_read": "0.0000011", + "input_cache_write": "0", + "min_prompt_tokens": 200000, + "prompt": "0.0000044" + } + ], + "prompt": "0.0000022", + "web_search": "0.01" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "top_k", + "stop", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock/us-west-2", + "throughput_last_30m": null, + "uptime_last_1d": 99.19354838709677, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.6", + "model_name": "SpaceXAI: Grok 4.6", + "name": "xAI | x-ai/grok-4.6-20260810", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.000001", + "overrides": [ + { + "completion": "0.000024", + "input_cache_read": "0.000002", + "min_prompt_tokens": 200000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/zdr/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.69496695475343, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 500000, + "latency_last_30m": null, + "max_completion_tokens": 450000, + "max_prompt_tokens": null, + "model_id": "x-ai/grok-4.6", + "model_name": "SpaceXAI: Grok 4.6", + "name": "xAI | x-ai/grok-4.6-20260810", + "pricing": { + "completion": "0.000012", + "discount": 0, + "input_cache_read": "0.000001", + "overrides": [ + { + "completion": "0.000024", + "input_cache_read": "0.000002", + "min_prompt_tokens": 200000, + "prompt": "0.000008" + } + ], + "prompt": "0.000004", + "web_search": "0.005" + }, + "provider_name": "xAI", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "xai/priority", + "throughput_last_30m": null, + "uptime_last_1d": 99.78260869565217, + "uptime_last_30m": null, + "uptime_last_5m": null + } + ], + "id": "x-ai/grok-4.6", + "name": "SpaceXAI: Grok 4.6" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1753471258, + "description": "GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter...", + "endpoints": [ + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 98304, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.5-air", + "model_name": "Z.ai: GLM 4.5 Air", + "name": "Novita | z-ai/glm-4.5-air", + "pricing": { + "completion": "0.00000085", + "discount": 0, + "input_cache_read": "0.000000025", + "prompt": "0.00000013" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 99.77065802179611, + "uptime_last_30m": 98.90829694323145, + "uptime_last_5m": 99.02912621359224 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.5-air", + "model_name": "Z.ai: GLM 4.5 Air", + "name": "SiliconFlow | z-ai/glm-4.5-air", + "pricing": { + "completion": "0.00000086", + "discount": 0, + "prompt": "0.00000014" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 98304, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.5-air", + "model_name": "Z.ai: GLM 4.5 Air", + "name": "Z.AI | z-ai/glm-4.5-air", + "pricing": { + "completion": "0.0000011", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.0000002" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.35355285961872, + "uptime_last_30m": 99.56521739130434, + "uptime_last_5m": 100 + } + ], + "id": "z-ai/glm-4.5-air", + "name": "Z.ai: GLM 4.5 Air" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1768833913, + "description": "As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency. It is further optimized for agentic coding use cases, strengthening coding capabilities, long-horizon task planning,...", + "endpoints": [ + { + "context_length": 128000, + "latency_last_30m": null, + "max_completion_tokens": 16384, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.7-flash", + "model_name": "Z.ai: GLM 4.7 Flash", + "name": "Venice | z-ai/glm-4.7-flash-20260119", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.00000001", + "prompt": "0.00000006" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 95.1683140941453, + "uptime_last_30m": 97.66899766899768, + "uptime_last_5m": 97.54098360655738 + }, + { + "context_length": 131072, + "latency_last_30m": null, + "max_completion_tokens": 117964, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.7-flash", + "model_name": "Z.ai: GLM 4.7 Flash", + "name": "Cloudflare | z-ai/glm-4.7-flash-20260119", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "prompt": "0.0000000605" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "structured_outputs", + "response_format", + "min_p", + "stop", + "logit_bias", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.20720438407113, + "uptime_last_30m": 99.84151646985705, + "uptime_last_5m": 99.75228658536585 + }, + { + "context_length": 200000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-4.7-flash", + "model_name": "Z.ai: GLM 4.7 Flash", + "name": "Novita | z-ai/glm-4.7-flash-20260119", + "pricing": { + "completion": "0.0000004", + "discount": 0, + "input_cache_read": "0.00000001", + "prompt": "0.00000007" + }, + "provider_name": "Novita", + "quantization": "bf16", + "status": -5, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/bf16", + "throughput_last_30m": null, + "uptime_last_1d": 61.25470349496377, + "uptime_last_30m": 41.51624548736462, + "uptime_last_5m": 42.16216216216216 + } + ], + "id": "z-ai/glm-4.7-flash", + "name": "Z.ai: GLM 4.7 Flash" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1770829182, + "description": "GLM-5 is Z.ai\u2019s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for expert developers, it delivers production-grade performance on large-scale programming tasks, rivaling leading...", + "endpoints": [ + { + "context_length": 198000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "StreamLake | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.00000192", + "discount": 0.4, + "input_cache_read": "0.00000012", + "prompt": "0.0000006" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "tool_choice", + "tools", + "max_tokens", + "temperature", + "top_p", + "stop", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.6152797552331, + "uptime_last_30m": 99.70781592403213, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "GMICloud | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.00000192", + "discount": 0.4, + "input_cache_read": "0.00000012", + "prompt": "0.0000006" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "tools", + "tool_choice", + "structured_outputs", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.48913519031647, + "uptime_last_30m": 98.81154499151104, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "Baidu | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.00000224", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000007" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "tool_choice", + "tools", + "response_format", + "structured_outputs", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed" + ], + "supports_implicit_caching": true, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.7803538743136, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 204800, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "SiliconFlow | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.00000255", + "discount": 0, + "input_cache_read": "0.0000002", + "prompt": "0.00000095" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.63663651802995, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "Amazon Bedrock | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "prompt": "0.000001" + }, + "provider_name": "Amazon Bedrock", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "tools", + "tool_choice", + "seed" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "amazon-bedrock", + "throughput_last_30m": null, + "uptime_last_1d": 91.63119489206446, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 198000, + "latency_last_30m": null, + "max_completion_tokens": 32000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "Venice | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "input_cache_read": "0.0000002", + "prompt": "0.000001" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "tools", + "tool_choice", + "response_format", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.82232976149325, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 202800, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "Novita | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "input_cache_read": "0.0000002", + "prompt": "0.000001" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.97475385003787, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5", + "model_name": "Z.ai: GLM 5", + "name": "Z.AI | z-ai/glm-5-20260211", + "pricing": { + "completion": "0.0000032", + "discount": 0, + "input_cache_read": "0.0000002", + "prompt": "0.000001" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.7449322056652, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "z-ai/glm-5", + "name": "Z.ai: GLM 5" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1775578025, + "description": "GLM-5.1 delivers a major leap in coding capability, with particularly significant gains in handling long-horizon tasks. Unlike previous models built around minute-level interactions, GLM-5.1 can work independently and continuously on...", + "endpoints": [ + { + "context_length": 200000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "StreamLake | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.000003036", + "discount": 0.31, + "input_cache_read": "0.0000001794", + "prompt": "0.000000966" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "tools", + "tool_choice", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.32166234088535, + "uptime_last_30m": 99.37849596022374, + "uptime_last_5m": 99.35064935064936 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 65535, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Chutes | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.00000308", + "discount": 0, + "input_cache_read": "0.000000098", + "prompt": "0.00000098" + }, + "provider_name": "Chutes", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "structured_outputs", + "tools", + "tool_choice", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "chutes/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 84.96967836032854, + "uptime_last_30m": 87.28121353558926, + "uptime_last_5m": 83.1896551724138 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 65536, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "DeepInfra | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000035", + "discount": 0, + "input_cache_read": "0.000000205", + "prompt": "0.00000105" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "tools", + "tool_choice", + "structured_outputs", + "logit_bias", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.62990805528248, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 204800, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "SiliconFlow | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.00000374", + "discount": 0, + "input_cache_read": "0.0000006", + "prompt": "0.00000119" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "tools", + "tool_choice", + "max_tokens" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.9436873521793, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Phala | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000042", + "discount": 0, + "input_cache_read": "0.0000006", + "prompt": "0.00000121" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "tool_choice", + "tools", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 83.74083129584352, + "uptime_last_30m": 87.01504354711005, + "uptime_last_5m": 87.05035971223022 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "AtlasCloud | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.00000396", + "discount": 0, + "input_cache_read": "0.000000234", + "prompt": "0.00000126" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.70510593312025, + "uptime_last_30m": 99.73045822102425, + "uptime_last_5m": 99.56989247311829 + }, + { + "context_length": 202745, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 169984, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Alibaba | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.00000418", + "discount": 0, + "input_cache_read": "0.000000247", + "prompt": "0.00000133" + }, + "provider_name": "Alibaba", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 204800, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Novita | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.00000138" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.90617988986334, + "uptime_last_30m": 99.87864077669903, + "uptime_last_5m": 99.05660377358491 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Nebius | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "prompt": "0.0000014" + }, + "provider_name": "Nebius", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "logit_bias", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": false, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nebius/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 84.22310416933384, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Baidu | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.66263674872945, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "GMICloud | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "logprobs", + "top_logprobs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.4700796745211, + "uptime_last_30m": 99.59239130434783, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Friendli | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Friendli", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "friendli", + "throughput_last_30m": null, + "uptime_last_1d": 99.98249197548876, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Z.AI | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k", + "response_format" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.90804245138335, + "uptime_last_30m": 99.62928637627432, + "uptime_last_5m": 100 + }, + { + "context_length": 200000, + "latency_last_30m": null, + "max_completion_tokens": 80000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.1", + "model_name": "Z.ai: GLM 5.1", + "name": "Venice | z-ai/glm-5.1-20260406", + "pricing": { + "completion": "0.0000044044", + "discount": 0.09, + "input_cache_read": "0.00000026026", + "prompt": "0.0000014014" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 89.78676609563738, + "uptime_last_30m": 94.79218828242364, + "uptime_last_5m": 53.63636363636364 + } + ], + "id": "z-ai/glm-5.1", + "name": "Z.ai: GLM 5.1" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1781631930, + "description": "GLM 5.2 is a large-scale reasoning model from Z.ai. It supports text input and output with a 1M-token context window, and is suited for long-horizon agent workflows, project-level software engineering,...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 163840, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "DeepInfra | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000018", + "discount": 0.25, + "input_cache_read": "0.000000105", + "prompt": "0.0000005625" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "logit_bias", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.67178214238179, + "uptime_last_30m": 98.59903381642512, + "uptime_last_5m": 98.56306546035125 + }, + { + "context_length": 202752, + "latency_last_30m": null, + "max_completion_tokens": 182476, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Ambient | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.000002", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.0000006" + }, + "provider_name": "Ambient", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "top_p", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "ambient/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.01142911117408, + "uptime_last_30m": 99.71098265895954, + "uptime_last_5m": 99.32885906040269 + }, + { + "context_length": 1024000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "StreamLake | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000020196", + "discount": 0.541, + "input_cache_read": "0.00000011934", + "prompt": "0.0000006426" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "tool_choice", + "tools", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "max_tokens", + "temperature", + "top_p", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.490258618789, + "uptime_last_30m": 99.76198571914315, + "uptime_last_5m": 99.74424552429667 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Novita | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000020416", + "discount": 0.536, + "input_cache_read": "0.00000012064", + "prompt": "0.0000006496" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.97006696870643, + "uptime_last_30m": 99.9970199070211, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "DigitalOcean | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000022", + "discount": 0, + "input_cache_read": "0.000000105", + "prompt": "0.0000007" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "tool_choice", + "tools", + "response_format", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 98.66027040601905, + "uptime_last_30m": 98.91348088531187, + "uptime_last_5m": 98.74281609195403 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "CoreWeave | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.00000242", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.00000076" + }, + "provider_name": "CoreWeave", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "coreweave/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.46572414926845, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "AtlasCloud | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.000002948", + "discount": 0.33, + "input_cache_read": "0.0000001742", + "prompt": "0.000000938" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.80322223588735, + "uptime_last_30m": 99.32885906040269, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 1048576, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Alibaba | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.000003036", + "discount": 0, + "input_cache_read": "0.0000001932", + "prompt": "0.000000966" + }, + "provider_name": "Alibaba", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.93556604480408, + "uptime_last_30m": 99.92010652463382, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "SiliconFlow | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.00000374", + "discount": 0.15, + "input_cache_read": "0.000000221", + "prompt": "0.00000119" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.99641787989663, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Phala | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.000003", + "discount": 0, + "input_cache_read": "0.00000022", + "prompt": "0.00000126" + }, + "provider_name": "Phala", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.80648145525946, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Inceptron | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000034506", + "discount": 0, + "input_cache_read": "0.0000002077", + "prompt": "0.000001269" + }, + "provider_name": "Inceptron", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "top_p", + "stop", + "frequency_penalty", + "logit_bias", + "parallel_tool_calls", + "presence_penalty", + "seed", + "response_format", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inceptron/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 97.40138124780522, + "uptime_last_30m": 93.68512110726644, + "uptime_last_5m": 36.486486486486484 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "BaseTen | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.97231575009353, + "uptime_last_30m": 99.95507637017072, + "uptime_last_5m": 100 + }, + { + "context_length": 512000, + "latency_last_30m": null, + "max_completion_tokens": 460800, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Together | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.6529778070392, + "uptime_last_30m": 99.67462039045553, + "uptime_last_5m": 99.52606635071089 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Baidu | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.96025602034891, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Wafer | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer", + "throughput_last_30m": null, + "uptime_last_1d": 99.14887262953562, + "uptime_last_30m": 99.81735159817352, + "uptime_last_5m": 99.84779299847793 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Fireworks | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.89379396028295, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "BaseTen | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.9141081871345, + "uptime_last_30m": 99.895178197065, + "uptime_last_5m": 100 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Venice | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Venice", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.69701890364408, + "uptime_last_30m": 98.37874352623282, + "uptime_last_5m": 99.2831541218638 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "GMICloud | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "structured_outputs", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.58209509658246, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Parasail | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Parasail", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.89822669681217, + "uptime_last_30m": 99.65753424657534, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Friendli | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Friendli", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "friendli", + "throughput_last_30m": null, + "uptime_last_1d": 99.91073783295803, + "uptime_last_30m": 99.90864799025579, + "uptime_last_5m": 100 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Cloudflare | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "min_p", + "stop", + "logit_bias", + "tools", + "structured_outputs", + "response_format", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.97508098679292, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Z.AI | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.81974980841672, + "uptime_last_30m": 99.91204925241864, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "BaseTen | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.95976025584005, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Fireworks | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks/fast-us", + "throughput_last_30m": null, + "uptime_last_1d": 99.93751394777951, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "BaseTen | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "structured_outputs", + "response_format", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.95976502776213, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Fireworks | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tool_choice", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.91642573183954, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Decart | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.0000072", + "discount": 0, + "input_cache_read": "0.00000042", + "prompt": "0.0000021" + }, + "provider_name": "Decart", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "decart/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.87155039242204, + "uptime_last_30m": 99.94005694590139, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Baidu | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.00000788", + "discount": 0, + "input_cache_read": "0.00000056", + "prompt": "0.00000225" + }, + "provider_name": "Baidu", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fast", + "throughput_last_30m": null, + "uptime_last_1d": 99.98470045680065, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": 1048576, + "model_id": "z-ai/glm-5.2", + "model_name": "Z.ai: GLM 5.2", + "name": "Alibaba | z-ai/glm-5.2-20260616", + "pricing": { + "completion": "0.00000726", + "discount": 0, + "input_cache_read": "0.000000462", + "prompt": "0.00000231" + }, + "provider_name": "Alibaba", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "top_k", + "frequency_penalty", + "stop", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba/fast", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + } + ], + "id": "z-ai/glm-5.2", + "name": "Z.ai: GLM 5.2" + }, + { + "architecture": { + "input_modalities": [ + "text" + ], + "instruct_type": null, + "modality": "text->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1787086655, + "description": "GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...", + "endpoints": [ + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Reka | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000297", + "discount": 0.25, + "input_cache_read": "0.0000001755", + "prompt": "0.0000008775" + }, + "provider_name": "Reka", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "max_tokens", + "stop", + "seed", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "reka/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.51336636963623, + "uptime_last_30m": 98.67525957751522, + "uptime_last_5m": 99.24242424242425 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "InferenceNet | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.000003", + "discount": 0, + "input_cache_read": "0.00000015", + "prompt": "0.0000009" + }, + "provider_name": "InferenceNet", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inference-net/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 96.69264172268159, + "uptime_last_30m": 93.33333333333333, + "uptime_last_5m": 94.73684210526315 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "DeepInfra | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.000003", + "discount": 0.25, + "input_cache_read": "0.00000015", + "prompt": "0.0000009" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 95.62755146356294, + "uptime_last_30m": 90.18691588785047, + "uptime_last_5m": 96.66666666666667 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Morph | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000283118", + "discount": 0.243, + "input_cache_read": "0.0000001479935", + "prompt": "0.00000090083" + }, + "provider_name": "Morph", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "stop", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "morph/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.57665712141416, + "uptime_last_30m": 99.66456195737963, + "uptime_last_5m": 99.6124031007752 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Phala | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000286", + "discount": 0.35, + "input_cache_read": "0.000000169", + "prompt": "0.00000091" + }, + "provider_name": "Phala", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala", + "throughput_last_30m": null, + "uptime_last_1d": 99.86010728683216, + "uptime_last_30m": 99.9666629625514, + "uptime_last_5m": 99.90909090909092 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Novita | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000286", + "discount": 0.35, + "input_cache_read": "0.000000169", + "prompt": "0.00000091" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.66536458614809, + "uptime_last_30m": 99.97576345128454, + "uptime_last_5m": 99.87886129618413 + }, + { + "context_length": 262124, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Io Net | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000308", + "discount": 0.3, + "input_cache_read": "0.000000182", + "prompt": "0.00000091" + }, + "provider_name": "Io Net", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "response_format", + "logprobs", + "top_logprobs", + "seed", + "stop", + "min_p", + "top_k", + "top_p", + "presence_penalty", + "frequency_penalty", + "repetition_penalty", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "io-net/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.34015849801027, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "GMICloud | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000033", + "discount": 0.25, + "input_cache_read": "0.000000195", + "prompt": "0.00000105" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.53634707733069, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 980000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Makora | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000042", + "discount": 0, + "input_cache_read": "0.00000019", + "prompt": "0.00000105" + }, + "provider_name": "Makora", + "quantization": "fp4", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "max_tokens", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "makora/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 94.9579831932773, + "uptime_last_30m": 89.2018779342723, + "uptime_last_5m": 72.97297297297297 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Inceptron | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000040805", + "discount": 0, + "input_cache_read": "0.0000001777", + "prompt": "0.0000010669" + }, + "provider_name": "Inceptron", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "max_tokens", + "top_p", + "stop", + "frequency_penalty", + "logit_bias", + "parallel_tool_calls", + "presence_penalty", + "seed", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inceptron/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.20070603237797, + "uptime_last_30m": 99.18950212273253, + "uptime_last_5m": 95.1492537313433 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "DigitalOcean | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.000003388", + "discount": 0, + "input_cache_read": "0.0000002002", + "prompt": "0.000001078" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 99.40524963863851, + "uptime_last_30m": 99.95085995085995, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "SiliconFlow | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000352", + "discount": 0.2, + "input_cache_read": "0.000000208", + "prompt": "0.00000112" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.67352082020734, + "uptime_last_30m": 100, + "uptime_last_5m": null + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Alibaba | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000374", + "discount": 0, + "input_cache_read": "0.000000238", + "prompt": "0.00000119" + }, + "provider_name": "Alibaba", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "presence_penalty", + "response_format", + "top_k", + "frequency_penalty", + "stop", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "alibaba", + "throughput_last_30m": null, + "uptime_last_1d": 99.91291182657395, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Decart | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000374", + "discount": 0.15, + "input_cache_read": "0.0000001955", + "prompt": "0.00000119" + }, + "provider_name": "Decart", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "decart/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.8767650834403, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Sail Research | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.000003871", + "discount": 0, + "input_cache_read": "0.000000225", + "prompt": "0.000001207" + }, + "provider_name": "Sail Research", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "frequency_penalty", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_k", + "top_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/us", + "throughput_last_30m": null, + "uptime_last_1d": 99.87184245320219, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Sail Research | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.000003871", + "discount": 0, + "input_cache_read": "0.000000225", + "prompt": "0.000001207" + }, + "provider_name": "Sail Research", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.91154600989579, + "uptime_last_30m": 99.88109393579072, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Friendli | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.00000396", + "discount": 0.1, + "input_cache_read": "0.000000234", + "prompt": "0.00000126" + }, + "provider_name": "Friendli", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "friendli", + "throughput_last_30m": null, + "uptime_last_1d": 99.99604756969453, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "AkashML | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000013" + }, + "provider_name": "AkashML", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "max_tokens", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "akashml/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.4286941580756, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Baidu | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Baidu", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baidu/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 100, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "BaseTen | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.77480803307738, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Mistral | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "Mistral", + "quantization": "nvfp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "response_format", + "structured_outputs", + "tools", + "logprobs", + "top_logprobs", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "mistral/nvfp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.81731136370313, + "uptime_last_30m": 99.94776704100288, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Crusoe | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Crusoe", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "min_p", + "repetition_penalty", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "crusoe/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 97.51246645860445, + "uptime_last_30m": 97.93195463642428, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Wafer | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer", + "throughput_last_30m": null, + "uptime_last_1d": 99.5315531777894, + "uptime_last_30m": 99.74072052401746, + "uptime_last_5m": 99.55947136563876 + }, + { + "context_length": 1000000, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Venice | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0.2, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 95.60499780797895, + "uptime_last_30m": 89.21161825726142, + "uptime_last_5m": null + }, + { + "context_length": 1048575, + "latency_last_30m": null, + "max_completion_tokens": 943717, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Together | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "logprobs", + "top_logprobs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 95.52432338152033, + "uptime_last_30m": 97.01492537313433, + "uptime_last_5m": 99.80456026058631 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Parasail | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.13321923480373, + "uptime_last_30m": 99.28343949044586, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Modal | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Modal", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "modal", + "throughput_last_30m": null, + "uptime_last_1d": 99.16040105270326, + "uptime_last_30m": 98.68573637417859, + "uptime_last_5m": 95.87554269175108 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "BaseTen | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000014", + "prompt": "0.0000014" + }, + "provider_name": "BaseTen", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.61865141272318, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Fireworks | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.9352602537798, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1310720, + "latency_last_30m": null, + "max_completion_tokens": 1179648, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Cloudflare | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "min_p", + "stop", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.90500316656112, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "AtlasCloud | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.9135097733956, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "Z.AI | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000044", + "discount": 0, + "input_cache_read": "0.00000026", + "prompt": "0.0000014" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.94034117758245, + "uptime_last_30m": 99.98813760379596, + "uptime_last_5m": 99.94350282485875 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "BaseTen | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 94.696289646918, + "uptime_last_30m": 98.71382636655949, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3", + "model_name": "Z.ai: GLM 5.3", + "name": "BaseTen | z-ai/glm-5.3-20260816", + "pricing": { + "completion": "0.0000066", + "discount": 0, + "input_cache_read": "0.00000021", + "prompt": "0.0000021" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 93.94828680203045, + "uptime_last_30m": 99.1869918699187, + "uptime_last_5m": 100 + } + ], + "id": "z-ai/glm-5.3", + "name": "Z.ai: GLM 5.3" + }, + { + "architecture": { + "input_modalities": [ + "text", + "image", + "video" + ], + "instruct_type": null, + "modality": "text+image+video->text", + "output_modalities": [ + "text" + ], + "tokenizer": "Other" + }, + "created": 1787752741, + "description": "GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...", + "endpoints": [ + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "DeepInfra | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000025", + "discount": 0.5, + "input_cache_read": "0.000000015", + "prompt": "0.000000075" + }, + "provider_name": "DeepInfra", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "top_k", + "seed", + "min_p", + "logit_bias", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "deepinfra/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 98.4672541837359, + "uptime_last_30m": 99.19301348662393, + "uptime_last_5m": 99.90128331688055 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "InferenceNet | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000028", + "discount": 0, + "input_cache_read": "0.00000002", + "prompt": "0.00000009" + }, + "provider_name": "InferenceNet", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "inference-net/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 95.82414561853946, + "uptime_last_30m": 95.3636627064619, + "uptime_last_5m": 96.89898679766657 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Relace | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000003", + "discount": 0, + "input_cache_read": "0.000000018", + "prompt": "0.00000009" + }, + "provider_name": "Relace", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "relace", + "throughput_last_30m": null, + "uptime_last_1d": 99.91881764404542, + "uptime_last_30m": 99.91729357728167, + "uptime_last_5m": 99.9481978840828 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Wafer | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000035", + "discount": 0, + "input_cache_read": "0.00000002", + "prompt": "0.0000001" + }, + "provider_name": "Wafer", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "max_tokens", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "structured_outputs", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "wafer", + "throughput_last_30m": null, + "uptime_last_1d": 99.64177910913446, + "uptime_last_30m": 99.88910451899086, + "uptime_last_5m": 99.80688767299645 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "GMICloud | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000035", + "discount": 0.3, + "input_cache_read": "0.000000021", + "prompt": "0.000000105" + }, + "provider_name": "GMICloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "seed", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "gmicloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.16248675379724, + "uptime_last_30m": 98.632587351433, + "uptime_last_5m": 99.35195852534562 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Novita | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000044", + "discount": 0.12, + "input_cache_read": "0.0000000264", + "prompt": "0.000000132" + }, + "provider_name": "Novita", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "repetition_penalty", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "novita/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.9788823227315, + "uptime_last_30m": 99.17274573212002, + "uptime_last_5m": 99.7232889089677 + }, + { + "context_length": 1024000, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "StreamLake | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.00000047", + "discount": 0.06, + "input_cache_read": "0.0000000282", + "prompt": "0.000000141" + }, + "provider_name": "StreamLake", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "response_format", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "streamlake/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.43191339162549, + "uptime_last_30m": 99.28259413959124, + "uptime_last_5m": 99.77511244377811 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Sail Research | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.000000475", + "discount": 0, + "input_cache_read": "0.0000000285", + "prompt": "0.0000001425" + }, + "provider_name": "Sail Research", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "frequency_penalty", + "max_tokens", + "presence_penalty", + "seed", + "stop", + "temperature", + "top_k", + "top_p", + "structured_outputs", + "tools", + "tool_choice", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/us", + "throughput_last_30m": null, + "uptime_last_1d": 98.75863176489482, + "uptime_last_30m": null, + "uptime_last_5m": null + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Sail Research | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.000000475", + "discount": 0, + "input_cache_read": "0.0000000285", + "prompt": "0.0000001425" + }, + "provider_name": "Sail Research", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "sail-research/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.06239577608358, + "uptime_last_30m": 95.64664760088974, + "uptime_last_5m": 99.07094594594594 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "BaseTen | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.32575678587565, + "uptime_last_30m": 99, + "uptime_last_5m": 98.46153846153847 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Near AI | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.000000035", + "prompt": "0.00000015" + }, + "provider_name": "Near AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "max_tokens", + "stop", + "seed", + "logit_bias", + "response_format", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "structured_outputs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "near-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.4279196815157, + "uptime_last_30m": 98.71362940275651, + "uptime_last_5m": 98.48484848484848 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Crusoe | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Crusoe", + "quantization": "fp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "min_p", + "repetition_penalty", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "crusoe/fp4", + "throughput_last_30m": null, + "uptime_last_1d": 92.9308860497951, + "uptime_last_30m": 96.67542213883678, + "uptime_last_5m": 96.81456200227532 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "CoreWeave | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000005", + "prompt": "0.00000015" + }, + "provider_name": "CoreWeave", + "quantization": "nvfp4", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "stop", + "seed", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "coreweave/nvfp4", + "throughput_last_30m": null, + "uptime_last_1d": 99.02471461819793, + "uptime_last_30m": 100, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "AtlasCloud | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "AtlasCloud", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "min_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "stop", + "seed", + "logit_bias", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "atlas-cloud/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 96.23044219150731, + "uptime_last_30m": 97.28108006750422, + "uptime_last_5m": 96.76190476190476 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Fireworks | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Fireworks", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "fireworks", + "throughput_last_30m": null, + "uptime_last_1d": 99.72229300184814, + "uptime_last_30m": 99.94747899159664, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Phala | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Phala", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "phala/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 97.15891037725858, + "uptime_last_30m": 98.50714054462391, + "uptime_last_5m": 99.19571045576407 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Friendli | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Friendli", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "seed", + "top_k", + "min_p", + "repetition_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "friendli", + "throughput_last_30m": null, + "uptime_last_1d": 99.60861293696186, + "uptime_last_30m": 99.85705534000408, + "uptime_last_5m": 100 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 262144, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "SiliconFlow | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "SiliconFlow", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "frequency_penalty", + "max_tokens", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "siliconflow/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.11789429380968, + "uptime_last_30m": 99.79533134196261, + "uptime_last_5m": 99.82578397212544 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "DigitalOcean | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "DigitalOcean", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "digitalocean", + "throughput_last_30m": null, + "uptime_last_1d": 95.17582313959717, + "uptime_last_30m": 99.29294329682517, + "uptime_last_5m": 99.59704499664204 + }, + { + "context_length": 1048575, + "latency_last_30m": null, + "max_completion_tokens": 943717, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Together | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Together", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "repetition_penalty", + "logit_bias", + "min_p", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "together", + "throughput_last_30m": null, + "uptime_last_1d": 99.60667144067482, + "uptime_last_30m": 96.67117306797279, + "uptime_last_5m": 99.73168768446472 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 235929, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Reka | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Reka", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "top_k", + "max_tokens", + "stop", + "seed", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "reka/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.60362977205274, + "uptime_last_30m": 99.8891813270536, + "uptime_last_5m": 99.91460290350128 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Parasail | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Parasail", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "frequency_penalty", + "presence_penalty", + "repetition_penalty", + "seed", + "stop", + "top_k", + "logit_bias", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "parasail/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.38606312139787, + "uptime_last_30m": 99.06553826003727, + "uptime_last_5m": 99.6291718170581 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "BaseTen | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "BaseTen", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": true, + "required": false + }, + "supports_voice_cloning": false, + "tag": "baseten/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.57860345888537, + "uptime_last_30m": 99.79475034537202, + "uptime_last_5m": 99.91039426523297 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Venice | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Venice", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "top_k", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "venice", + "throughput_last_30m": null, + "uptime_last_1d": 96.73009047020163, + "uptime_last_30m": 96.62750217580505, + "uptime_last_5m": 97.36842105263158 + }, + { + "context_length": 262144, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Io Net | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Io Net", + "quantization": "fp8", + "status": -2, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "tools", + "tool_choice", + "logprobs", + "top_logprobs", + "seed", + "stop", + "min_p", + "top_k", + "top_p", + "presence_penalty", + "frequency_penalty", + "repetition_penalty", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "io-net/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.28830340661185, + "uptime_last_30m": 87.07440100882724, + "uptime_last_5m": 100 + }, + { + "context_length": 1310720, + "latency_last_30m": null, + "max_completion_tokens": 1179648, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Cloudflare | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Cloudflare", + "quantization": "unknown", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "top_k", + "seed", + "repetition_penalty", + "frequency_penalty", + "presence_penalty", + "min_p", + "stop", + "logit_bias", + "logprobs", + "top_logprobs", + "response_format", + "structured_outputs", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "cloudflare", + "throughput_last_30m": null, + "uptime_last_1d": 99.17520207549151, + "uptime_last_30m": 98.12792511700468, + "uptime_last_5m": 99.67105263157895 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 131072, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Z.AI | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000005", + "discount": 0, + "input_cache_read": "0.00000003", + "prompt": "0.00000015" + }, + "provider_name": "Z.AI", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "tools", + "tool_choice", + "top_k", + "response_format", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "z-ai/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.12314100423923, + "uptime_last_30m": 99.66922204374339, + "uptime_last_5m": 99.90490680867251 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 128000, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "NextBit | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000006", + "discount": 0, + "input_cache_read": "0.000000036", + "prompt": "0.00000018" + }, + "provider_name": "NextBit", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "max_tokens", + "temperature", + "top_p", + "stop", + "frequency_penalty", + "presence_penalty", + "response_format", + "structured_outputs", + "repetition_penalty", + "seed", + "tools", + "tool_choice", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": true, + "function": true, + "none": true, + "required": true + }, + "supports_voice_cloning": false, + "tag": "nextbit/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 98.66562257263926, + "uptime_last_30m": 99.85714285714286, + "uptime_last_5m": 99.83258928571429 + }, + { + "context_length": 1048576, + "latency_last_30m": null, + "max_completion_tokens": 943718, + "max_prompt_tokens": null, + "model_id": "z-ai/glm-5.3-flash", + "model_name": "Z.ai: GLM 5.3 Flash", + "name": "Modal | z-ai/glm-5.3-flash-20260826", + "pricing": { + "completion": "0.0000015", + "discount": 0, + "input_cache_read": "0.00000009", + "prompt": "0.00000045" + }, + "provider_name": "Modal", + "quantization": "fp8", + "status": 0, + "supported_parameters": [ + "reasoning", + "include_reasoning", + "temperature", + "top_p", + "stop", + "max_tokens", + "response_format", + "structured_outputs", + "tools", + "reasoning_effort" + ], + "supports_implicit_caching": false, + "supports_tool_choice": { + "auto": false, + "function": false, + "none": false, + "required": false + }, + "supports_voice_cloning": false, + "tag": "modal/fp8", + "throughput_last_30m": null, + "uptime_last_1d": 99.66626998723413, + "uptime_last_30m": 99.71842344941703, + "uptime_last_5m": 99.7044418172606 + } + ], + "id": "z-ai/glm-5.3-flash", + "name": "Z.ai: GLM 5.3 Flash" + } + ], + "requested_models": [ + "anthropic/claude-fable-5.1", + "deepseek/deepseek-chat-v3-0324", + "deepseek/deepseek-chat-v3.1", + "deepseek/deepseek-v3.2", + "deepseek/deepseek-v3.2-exp", + "deepseek/deepseek-v4-flash", + "deepseek/deepseek-v4-flash-0731", + "deepseek/deepseek-v4.1-flash", + "google/gemini-2.5-flash", + "google/gemini-2.5-flash-lite", + "google/gemini-3-flash-preview", + "google/gemini-3.1-flash-lite", + "google/gemini-3.1-flash-lite-preview", + "google/gemini-3.5-flash", + "google/gemini-3.5-flash-lite", + "google/gemini-3.6-flash", + "google/gemini-3.7-flash", + "google/gemini-3.8-flash", + "meta/muse-glimmer-30b", + "meta/muse-spark-1.3", + "moonshotai/kimi-k2.6", + "moonshotai/kimi-k3", + "openai/gpt-3.5-turbo-0613", + "openai/gpt-3.5-turbo-16k", + "openai/gpt-4.1", + "openai/gpt-4.1-mini", + "openai/gpt-4.1-nano", + "openai/gpt-4o", + "openai/gpt-4o-2024-08-06", + "openai/gpt-4o-2024-11-20", + "openai/gpt-4o-mini", + "openai/gpt-4o-mini-2024-07-18", + "openai/gpt-5", + "openai/gpt-5-mini", + "openai/gpt-5-nano", + "openai/gpt-5.1", + "openai/gpt-5.2", + "openai/gpt-5.4", + "openai/gpt-5.4-mini", + "openai/gpt-5.4-nano", + "openai/gpt-5.6-luna", + "openai/gpt-5.6-sol", + "openai/gpt-5.6-terra", + "openai/gpt-6-astra", + "openai/gpt-oss-120b", + "qwen/qwen-2.5-72b-instruct", + "qwen/qwen-2.5-7b-instruct", + "qwen/qwen-plus", + "qwen/qwen-plus-2025-07-28", + "qwen/qwen2.5-vl-72b-instruct", + "qwen/qwen3-14b", + "qwen/qwen3-235b-a22b", + "qwen/qwen3-235b-a22b-2507", + "qwen/qwen3-30b-a3b", + "qwen/qwen3-30b-a3b-instruct-2507", + "qwen/qwen3-32b", + "qwen/qwen3-8b", + "qwen/qwen3-coder", + "qwen/qwen3-coder-30b-a3b-instruct", + "qwen/qwen3-coder-next", + "qwen/qwen3-coder-plus", + "qwen/qwen3-max-thinking", + "qwen/qwen3-next-80b-a3b-instruct", + "qwen/qwen3-vl-235b-a22b-instruct", + "qwen/qwen3-vl-30b-a3b-instruct", + "qwen/qwen3-vl-32b-instruct", + "qwen/qwen3-vl-8b-instruct", + "qwen/qwen3.5-122b-a10b", + "qwen/qwen3.5-27b", + "qwen/qwen3.5-35b-a3b", + "qwen/qwen3.5-397b-a17b", + "qwen/qwen3.5-9b", + "qwen/qwen3.5-plus-02-15", + "qwen/qwen3.5-plus-20260420", + "qwen/qwen3.6-27b", + "qwen/qwen3.6-35b-a3b", + "qwen/qwen3.6-flash", + "qwen/qwen3.6-max-preview", + "qwen/qwen3.6-plus", + "qwen/qwen3.7-flash", + "qwen/qwen3.7-max", + "qwen/qwen3.7-plus", + "qwen/qwen3.8-2.4t-a95b", + "qwen/qwen3.8-27b", + "qwen/qwen3.8-flash", + "thinkingmachines/inkling", + "x-ai/grok-4.20", + "x-ai/grok-4.3", + "x-ai/grok-4.5", + "x-ai/grok-4.6", + "z-ai/glm-4.5-air", + "z-ai/glm-4.7-flash", + "z-ai/glm-5", + "z-ai/glm-5.1", + "z-ai/glm-5.2", + "z-ai/glm-5.3", + "z-ai/glm-5.3-flash" + ], + "schema": 2, + "scope": "all exact model IDs in canonical wvs-score-all-options-v1 cache, including unpublished complete cache entries", + "source": "OpenRouter public per-model endpoint API" +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/manifest.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/manifest.json new file mode 100644 index 0000000..f5054e7 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/manifest.json @@ -0,0 +1,673 @@ +{ + "catalog_sha256": "940edd61d8103a51322710d8a6baee7f698b64b2cf033a933f0321e15c536fd2", + "cells_per_model": 4, + "created_utc": "2026-09-18T11:40:01.832011+00:00", + "endpoint_catalog_sha256": "b7d824d27792b6f237d76ce8d1237780837d953943db77a7214fc6b034a7ae26", + "eval_version": "wvs-gemini-flash-rubric-v1", + "hard_pilot_stop_usd": "20", + "headroom_below_hard_stop_usd": "3.77267200", + "items": 12, + "max_attempts_per_request": 3, + "merge_reversed_into_primary": false, + "models": [ + { + "advertised_quantization": "unknown", + "calls": 288, + "cells": [ + { + "max_tokens": 2048, + "name": "normal_minimum", + "protocol_id": "e0d26259ddccd08fdba6baedd51acbb873dd0810e916a394be61b18ad5ae422a", + "rating_rubric": "normal", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3-flash-preview/normal_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "normal_high", + "protocol_id": "80307718e12e6ef071350f0a8e04e61de144eb35c61b624b8b2e3bc1d1945cc9", + "rating_rubric": "normal", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3-flash-preview/normal_high.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_minimum", + "protocol_id": "152844599eae3e1c4c8c40130bc7f68f50d996da0818a9a331d31b4f1107b672", + "rating_rubric": "reversed", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3-flash-preview/reversed_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_high", + "protocol_id": "9c2cef1728b21904289627949f2cb216807570a0a99198ae8b20a03cfea0397d", + "rating_rubric": "reversed", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3-flash-preview/reversed_high.jsonl" + } + ], + "created": 1765987078, + "id": "google/gemini-3-flash-preview", + "input_usd_per_million": "0.5000000", + "minimum_reasoning": "minimal", + "output_usd_per_million": "3.000000", + "paired_seed_schedule": [ + 1163952736, + 1170119667, + 2035573066, + 232577473, + 1683876624, + 235742503, + 57970632, + 678890407, + 1231480144, + 183286478, + 445597396, + 124377117, + 629484651, + 518426933, + 2071960594, + 618227503, + 1651495477, + 1008083987, + 158351131, + 168854583, + 2042985260, + 819696340, + 608320465, + 1527722754, + 203525376, + 156396839, + 1310326177, + 519158083, + 1777747065, + 27622176, + 1172729336, + 1371968231, + 1799582183, + 1760086890, + 1934536107, + 1642381013, + 887146042, + 1758492626, + 169490660, + 867622290, + 1799182150, + 479856359, + 1002977212, + 190582054, + 1855675895, + 1470637308, + 419372111, + 1315039314, + 17206795, + 617241580, + 2123813183, + 78370729, + 2083460461, + 115192274, + 494620397, + 572413235, + 677340727, + 1153795536, + 1222892914, + 1130618781, + 1896231879, + 1067456760, + 129880745, + 1961307608, + 1628746868, + 1524161661, + 634359330, + 1295199692, + 1019390209, + 1495255605, + 2021003453, + 501957876 + ], + "provider_endpoint": "google-ai-studio", + "runtime_initial_request_bound_usd": "2.0643840" + }, + { + "advertised_quantization": "unknown", + "calls": 288, + "cells": [ + { + "max_tokens": 2048, + "name": "normal_minimum", + "protocol_id": "5c0f17c2ba7fdb7b5616a5d1907ad6b657a4180dc5d783c653f57d94debef777", + "rating_rubric": "normal", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.5-flash/normal_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "normal_high", + "protocol_id": "998703570404688e41278cc8e9eb50aac4b38185f2a18b061f1e710564a40147", + "rating_rubric": "normal", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.5-flash/normal_high.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_minimum", + "protocol_id": "582f12aa78a9fad7c8f921ba27019ce6ba8b08b7100d4d5892a1c12f7c088c91", + "rating_rubric": "reversed", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.5-flash/reversed_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_high", + "protocol_id": "f3e2146f8613fc787a2c949051be2bf4f91972157ec5a2a852a65e94ef902815", + "rating_rubric": "reversed", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.5-flash/reversed_high.jsonl" + } + ], + "created": 1779193800, + "id": "google/gemini-3.5-flash", + "input_usd_per_million": "1.5000000", + "minimum_reasoning": "minimal", + "output_usd_per_million": "9.000000", + "paired_seed_schedule": [ + 1248808999, + 1130076528, + 127109785, + 1583457001, + 1355118102, + 1964788436, + 859880036, + 1662291375, + 1978299614, + 1630597822, + 327086889, + 214904754, + 895639426, + 160461167, + 1129826421, + 450298793, + 1355440748, + 1275111749, + 991324399, + 1696942014, + 547236783, + 1663182698, + 2104409755, + 1758476120, + 874531288, + 24483151, + 959677948, + 1147026597, + 1046590322, + 954734840, + 984925420, + 795650251, + 1375694599, + 608850335, + 184050643, + 2117464855, + 1008371759, + 549016040, + 456297439, + 102113382, + 1567517018, + 1978075073, + 1331081413, + 1065105221, + 1288270755, + 464227865, + 837038729, + 1482029053, + 1970697406, + 1543875046, + 1585450653, + 2095546870, + 2028897412, + 261729936, + 1275645584, + 515424394, + 1762341000, + 328052941, + 925967965, + 772020185, + 1415060343, + 719448036, + 1245354653, + 64796124, + 536996806, + 110003216, + 1259134782, + 1338283125, + 752914028, + 721340567, + 838155882, + 352115216 + ], + "provider_endpoint": "google-ai-studio", + "runtime_initial_request_bound_usd": "6.1931520" + }, + { + "advertised_quantization": "unknown", + "calls": 288, + "cells": [ + { + "max_tokens": 2048, + "name": "normal_minimum", + "protocol_id": "ee6923acc4e8a4b72b72a8f35899b4f3ff440f134ee8b88676ca83d02f742763", + "rating_rubric": "normal", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.6-flash/normal_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "normal_high", + "protocol_id": "908521e54c5896fb9daf6126670a9c0e9ce52efe095d6115df45868a4f973c34", + "rating_rubric": "normal", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.6-flash/normal_high.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_minimum", + "protocol_id": "6adfdd902f37a63a0980730c43ad977c2cff9248469b7b6602f0d5c80254c285", + "rating_rubric": "reversed", + "reasoning": { + "effort": "minimal" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.6-flash/reversed_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_high", + "protocol_id": "384c0dbfbdb1bbf7caf981c2815e73407507a0e14ef7da8410fecc9288e56089", + "rating_rubric": "reversed", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.6-flash/reversed_high.jsonl" + } + ], + "created": 1784646733, + "id": "google/gemini-3.6-flash", + "input_usd_per_million": "0.75000000", + "minimum_reasoning": "minimal", + "output_usd_per_million": "3.75000000", + "paired_seed_schedule": [ + 108817882, + 1676038061, + 1275187563, + 92317517, + 483723645, + 1580919723, + 358338923, + 1595687570, + 2083342467, + 831892180, + 1774790740, + 1631244917, + 1511547344, + 414117051, + 1596860634, + 1878760624, + 342966582, + 400837473, + 1105373711, + 1481631090, + 1052779020, + 1270143699, + 1327202473, + 1175044801, + 1883430981, + 98309661, + 432413791, + 1407149942, + 1967376545, + 1945945222, + 100031864, + 12779760, + 282054530, + 1215416239, + 1755043759, + 1540478859, + 1150657313, + 751055404, + 1279377459, + 1732314870, + 746811485, + 1599265935, + 1234148501, + 813713317, + 923932206, + 1620867402, + 917315350, + 363098115, + 1332703528, + 1311003049, + 443310493, + 322303696, + 1262849871, + 2145390355, + 860175181, + 1261049515, + 1935856055, + 1450813262, + 1752864454, + 1840245421, + 118834715, + 1181444103, + 1608361040, + 1028416984, + 1156366144, + 1441136226, + 1604124651, + 1816779536, + 686683234, + 2004819118, + 1715971918, + 100473306 + ], + "provider_endpoint": "google-ai-studio", + "runtime_initial_request_bound_usd": "2.65420800" + }, + { + "advertised_quantization": "unknown", + "calls": 288, + "cells": [ + { + "max_tokens": 2048, + "name": "normal_minimum", + "protocol_id": "4abfb1e17b33a1ff2dd7057006d366930997829637c0faee188bf56a16a292f0", + "rating_rubric": "normal", + "reasoning": { + "effort": "low" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.7-flash/normal_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "normal_high", + "protocol_id": "8a18fcef54ac12b7cabfbc4e66bec22f207633f4f395e0419cb060ad55524c8e", + "rating_rubric": "normal", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.7-flash/normal_high.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_minimum", + "protocol_id": "818ffecb3ac5e9519591d194d80c8233292fc5e8cf4fb670ec7b2d871c54fa35", + "rating_rubric": "reversed", + "reasoning": { + "effort": "low" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.7-flash/reversed_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_high", + "protocol_id": "93085f421a57a6625d7d40be7db3251d0347c9ad10d2bc0c754357075cd258de", + "rating_rubric": "reversed", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.7-flash/reversed_high.jsonl" + } + ], + "created": 1786640581, + "id": "google/gemini-3.7-flash", + "input_usd_per_million": "0.75000000", + "minimum_reasoning": "low", + "output_usd_per_million": "3.75000000", + "paired_seed_schedule": [ + 910621187, + 1035003559, + 1288844299, + 1947833639, + 447300522, + 1381453150, + 1569917861, + 695487735, + 1713188587, + 2048368253, + 1682009709, + 469705899, + 1704273714, + 698854731, + 616801448, + 1613696824, + 123761298, + 167556361, + 463575904, + 394493140, + 568055740, + 1054086786, + 1593919371, + 231445895, + 1227309383, + 1306310380, + 494375025, + 812250409, + 1619513880, + 481598774, + 2010187172, + 1895470904, + 1936884259, + 2068659266, + 968460531, + 1624286125, + 1794979195, + 1767632927, + 1172721448, + 411453953, + 1957436835, + 1057643472, + 770645601, + 1701012062, + 1469726925, + 615140549, + 1299790767, + 150924191, + 848073898, + 1200529624, + 1214366506, + 850842566, + 549948318, + 2024856341, + 1759859358, + 736670162, + 1734999180, + 853445306, + 1811586779, + 1638035411, + 2005268256, + 943232658, + 1650802609, + 1893994865, + 500344098, + 1883701642, + 1993628006, + 335533332, + 766847249, + 754505371, + 305407183, + 1128794051 + ], + "provider_endpoint": "google-ai-studio", + "runtime_initial_request_bound_usd": "2.65420800" + }, + { + "advertised_quantization": "unknown", + "calls": 288, + "cells": [ + { + "max_tokens": 2048, + "name": "normal_minimum", + "protocol_id": "96474f9ed9017e6ae81c6e7214dab7e122d263ebc5c54abb6a74acaa120b4254", + "rating_rubric": "normal", + "reasoning": { + "effort": "low" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.8-flash/normal_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "normal_high", + "protocol_id": "39c670a238050ddeed2e6c1568285af45f1667562f07479ae3a64ea429f5d2a3", + "rating_rubric": "normal", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.8-flash/normal_high.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_minimum", + "protocol_id": "d35083caa6744676f4b483550d62f650a6121b8a02c55ede8ddd34c58f6fcff2", + "rating_rubric": "reversed", + "reasoning": { + "effort": "low" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.8-flash/reversed_minimum.jsonl" + }, + { + "max_tokens": 2048, + "name": "reversed_high", + "protocol_id": "2bf10e1d6fd2a0dc7aef8538423b966f3293ad2b881234f8c71e87b696849208", + "rating_rubric": "reversed", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/records/google__gemini-3.8-flash/reversed_high.jsonl" + } + ], + "created": 1788362056, + "id": "google/gemini-3.8-flash", + "input_usd_per_million": "0.75000000", + "minimum_reasoning": "low", + "output_usd_per_million": "3.75000000", + "paired_seed_schedule": [ + 391353894, + 1578387018, + 1550504460, + 789210169, + 1446327476, + 408191264, + 1758622650, + 246218892, + 1056135751, + 36329443, + 177689873, + 1356198839, + 1958602544, + 375292404, + 1391359247, + 1207351073, + 2009545550, + 1519001559, + 1899590285, + 1519851461, + 1888002127, + 2145028215, + 1285225754, + 2058839674, + 1772943871, + 133734367, + 646723797, + 894669559, + 389997562, + 159118852, + 1183875049, + 410617310, + 208666835, + 2025528617, + 196246283, + 1058763755, + 2091314350, + 1039466951, + 844676764, + 630844192, + 2106446312, + 1342262254, + 1370808870, + 380768268, + 1488261277, + 868255277, + 2032084613, + 1322383633, + 712786726, + 1849425665, + 1300055292, + 1749878781, + 1542269193, + 1849576459, + 230515792, + 1391622440, + 1196282919, + 178607872, + 1460316783, + 183004564, + 925546222, + 1471520620, + 1803874068, + 1709075211, + 14371788, + 1330624040, + 1856301928, + 1302970835, + 57568791, + 1140320448, + 614073330, + 1475936766 + ], + "provider_endpoint": "google-ai-studio", + "runtime_initial_request_bound_usd": "2.65420800" + } + ], + "not_published": true, + "paid_smoke_bound_usd": "0.0071680", + "paid_smoke_calls": 1, + "panel_calls": 1440, + "panel_initial_request_bound_usd": "16.22016000", + "panel_plus_smoke_bound_usd": "16.22732800", + "provider": { + "allow_fallbacks": false, + "only": [ + "google-ai-studio" + ], + "require_parameters": true + }, + "quantization_audit": "No quantization-stratified variance is identifiable. Exact requires an explicit saved response quantization; current endpoint joins are ambiguous rather than historical route evidence.", + "request_metadata": true, + "request_timeout_seconds": 400, + "reverse_transform": "6 - rating; raw and transformed samples retained separately", + "runtime_input_token_reserve_per_initial_request": 2048, + "samples_per_item_per_cell": 6, + "schema": 1, + "structured_output": true, + "temperature": 1.0, + "transport_timeout_seconds": 120, + "z_scaling": false +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.json new file mode 100644 index 0000000..1cbff5d --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.json @@ -0,0 +1,33 @@ +{ + "durable_response_field": { + "endpoints": { + "available": [ + { + "model": "fake-release", + "provider": "Google AI Studio", + "selected": true + } + ] + }, + "requested": "fake" + }, + "failed_retry_conservative_charge_usd": "0.0071680", + "request_header": "X-OpenRouter-Metadata: enabled", + "rescue_seed": "seed 23 present in both initial and rescue payloads", + "rescue_seed_records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/rescue_seed_capture_smoke.jsonl", + "resume": "one completed request reused without another call", + "retry": "one retryable timeout preserved, charged conservatively, then completed", + "reverse_transform": { + "raw": [ + 1.0, + 5.0 + ], + "transformed": [ + 5.0, + 1.0 + ] + }, + "route_mismatch": "failed inline after charging actual cost and saving the full response", + "route_mismatch_record": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/route_mismatch_capture_smoke.json", + "status": "passed" +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.jsonl b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.jsonl new file mode 100644 index 0000000..83c716c --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/metadata_capture_smoke.jsonl @@ -0,0 +1,11 @@ +{"event": "run_started", "items": [{"id": "test", "n": 2, "options": ["a", "b"], "question": "test"}], "planned_requests": 1, "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.540792+00:00", "run_id": "20260918T114624Z_d7c60b288222", "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 512, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "reversed", "reasoning": {"effort": "minimal"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "item_id": "test", "model": "fake", "payload": {"max_tokens": 512, "messages": [{"content": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}], "model": "fake", "n": 1, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "minimal"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1"], "type": "object"}, "strict": true}, "type": "json_schema"}, "seed": 1, "temperature": 1.0}, "phase": "initial", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.587911+00:00", "request_id": "20260918T114624Z_d7c60b288222_000", "run_id": "20260918T114624Z_d7c60b288222", "sample": 0, "seed": 1, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 512, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "reversed", "reasoning": {"effort": "minimal"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "test", "model": "fake", "phase": "initial", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:46:24.629425+00:00", "request_id": "20260918T114624Z_d7c60b288222_000", "response": {"choices": [{"finish_reason": "stop", "message": {"content": "{\"0\": 1, \"1\": 5}"}}], "id": "fake", "model": "fake", "openrouter_metadata": {"endpoints": {"available": [{"model": "fake-release", "provider": "Google AI Studio", "selected": true}]}, "requested": "fake"}, "provider": "Google AI Studio", "usage": {"cost": 0}}, "run_id": "20260918T114624Z_d7c60b288222", "sample": 0, "seed": 1, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 512, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "reversed", "reasoning": {"effort": "minimal"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1], "structured_output": true, "temperature": 1.0}, "usage": {"cost": 0}} +{"event": "answer_parsed", "item_id": "test", "model": "fake", "parsed": true, "presented_order": [0, 1], "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.679849+00:00", "request_id": "20260918T114624Z_d7c60b288222_000", "rescued": false, "run_id": "20260918T114624Z_d7c60b288222", "sample": 0, "text": "{\"0\": 1, \"1\": 5}"} +{"event": "item_result", "failed_samples": 0, "id": "test", "model": "fake", "n_samples": 1, "p": [0.8333333333333334, 0.16666666666666666], "p_samples": [[0.8333333333333334, 0.16666666666666666]], "pmass_allowed": 1.0, "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "rating_rubric": "reversed", "rating_samples_raw": [[1.0, 5.0]], "rating_samples_transformed": [[5.0, 1.0]], "recorded_at_utc": "2026-09-18T11:46:24.712899+00:00", "rescued_samples": 0, "run_id": "20260918T114624Z_d7c60b288222", "texts": ["{\"0\": 1, \"1\": 5}"], "valid_samples": 1} +{"event": "run_finished", "failed_samples": 0, "model": "fake", "planned_requests": 1, "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.746152+00:00", "rescued_samples": 0, "run_id": "20260918T114624Z_d7c60b288222", "valid_samples": 1} +{"event": "run_started", "items": [{"id": "test", "n": 2, "options": ["a", "b"], "question": "test"}], "planned_requests": 1, "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.779863+00:00", "run_id": "20260918T114624Z_d7c60b288222", "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 512, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "reversed", "reasoning": {"effort": "minimal"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": true, "seed_schedule": [1], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_reused", "item_id": "test", "model": "fake", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.813143+00:00", "request_id": "20260918T114624Z_d7c60b288222_000", "run_id": "20260918T114624Z_d7c60b288222", "sample": 0, "seed": 1, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 512, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "reversed", "reasoning": {"effort": "minimal"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": true, "seed_schedule": [1], "structured_output": true, "temperature": 1.0}, "source_run_id": "20260918T114624Z_d7c60b288222"} +{"event": "answer_parsed", "item_id": "test", "model": "fake", "parsed": true, "presented_order": [0, 1], "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.863271+00:00", "request_id": "20260918T114624Z_d7c60b288222_000", "rescued": false, "run_id": "20260918T114624Z_d7c60b288222", "sample": 0, "text": "{\"0\": 1, \"1\": 5}"} +{"event": "item_result", "failed_samples": 0, "id": "test", "model": "fake", "n_samples": 1, "p": [0.8333333333333334, 0.16666666666666666], "p_samples": [[0.8333333333333334, 0.16666666666666666]], "pmass_allowed": 1.0, "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly endorse) to 5 (strongly reject):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "rating_rubric": "reversed", "rating_samples_raw": [[1.0, 5.0]], "rating_samples_transformed": [[5.0, 1.0]], "recorded_at_utc": "2026-09-18T11:46:24.905013+00:00", "rescued_samples": 0, "run_id": "20260918T114624Z_d7c60b288222", "texts": ["{\"0\": 1, \"1\": 5}"], "valid_samples": 1} +{"event": "run_finished", "failed_samples": 0, "model": "fake", "planned_requests": 1, "protocol_id": "d7c60b28822215e0b812f81c2e77794690bc971d44a3b5eba2d44245dc9c6755", "recorded_at_utc": "2026-09-18T11:46:24.946569+00:00", "rescued_samples": 0, "run_id": "20260918T114624Z_d7c60b288222", "valid_samples": 1} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.json new file mode 100644 index 0000000..7e6880b --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.json @@ -0,0 +1,273 @@ +{ + "budget": { + "conservative_spent_usd": "0.0053585", + "failed_phases_charged_at_bound": 0, + "provider_reported_spent_usd": "0.0053585", + "reserved_usd": "0E-7" + }, + "cell": "normal_high", + "global_accounting": { + "external_observed_usd": { + "pilot/gemini-flash-rubric-smoke/20260918T111625.452461Z": "0.0053585" + }, + "matching_reservations": {}, + "total_external_observed_usd": "0.0053585" + }, + "model": "google/gemini-3-flash-preview", + "parse_valid": true, + "phases": [ + { + "content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3,", + "finish_reason": "length", + "phase": "initial", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", + "type": "reasoning.text" + } + ], + "reasoning_text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", + "response": { + "choices": [ + { + "finish_reason": "length", + "index": 0, + "logprobs": null, + "message": { + "content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3,", + "reasoning": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", + "type": "reasoning.text" + } + ], + "refusal": null, + "role": "assistant" + }, + "native_finish_reason": "MAX_TOKENS" + } + ], + "created": 1789730186, + "id": "gen-1789730186-UoVJdYDpZct760XocC0C", + "model": "google/gemini-3-flash-preview", + "object": "chat.completion", + "openrouter_metadata": { + "attempt": 1, + "endpoints": { + "available": [ + { + "model": "google/gemini-3-flash-preview-20251217", + "provider": "Google AI Studio", + "selected": true + } + ], + "total": 6 + }, + "generation_time": 5892, + "is_byok": false, + "region": "PER", + "requested": "google/gemini-3-flash-preview", + "strategy": "direct", + "summary": "available=1, selected=Google AI Studio" + }, + "provider": "Google AI Studio", + "service_tier": "default", + "system_fingerprint": null, + "usage": { + "completion_tokens": 1009, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 979 + }, + "cost": 0.00311, + "cost_details": { + "upstream_inference_completions_cost": 0.003027, + "upstream_inference_cost": 0.00311, + "upstream_inference_prompt_cost": 8.3e-05 + }, + "is_byok": false, + "prompt_tokens": 166, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1175 + } + }, + "response_id": "gen-1789730186-UoVJdYDpZct760XocC0C", + "route": { + "advertised_quantization": "unknown", + "endpoint_tag": "google-ai-studio", + "requested_model": "google/gemini-3-flash-preview", + "response_model": "google/gemini-3-flash-preview", + "selected_provider": "Google AI Studio", + "selected_release_slug": "google/gemini-3-flash-preview-20251217" + }, + "usage": { + "completion_tokens": 1009, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 979 + }, + "cost": 0.00311, + "cost_details": { + "upstream_inference_completions_cost": 0.003027, + "upstream_inference_cost": 0.00311, + "upstream_inference_prompt_cost": 8.3e-05 + }, + "is_byok": false, + "prompt_tokens": 166, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1175 + } + }, + { + "content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}", + "finish_reason": "stop", + "phase": "rescue", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", + "type": "reasoning.text" + } + ], + "reasoning_text": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", + "response": { + "choices": [ + { + "finish_reason": "stop", + "index": 0, + "logprobs": null, + "message": { + "content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}", + "reasoning": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", + "type": "reasoning.text" + } + ], + "refusal": null, + "role": "assistant" + }, + "native_finish_reason": "STOP" + } + ], + "created": 1789730193, + "id": "gen-1789730193-mKoXCXzrJe42P1k6Mhw6", + "model": "google/gemini-3-flash-preview", + "object": "chat.completion", + "openrouter_metadata": { + "attempt": 1, + "endpoints": { + "available": [ + { + "model": "google/gemini-3-flash-preview-20251217", + "provider": "Google AI Studio", + "selected": true + } + ], + "total": 6 + }, + "generation_time": 4038, + "is_byok": false, + "region": "PER", + "requested": "google/gemini-3-flash-preview", + "strategy": "direct", + "summary": "available=1, selected=Google AI Studio" + }, + "provider": "Google AI Studio", + "service_tier": "default", + "system_fingerprint": null, + "usage": { + "completion_tokens": 662, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 602 + }, + "cost": 0.0022485, + "cost_details": { + "upstream_inference_completions_cost": 0.001986, + "upstream_inference_cost": 0.0022485, + "upstream_inference_prompt_cost": 0.0002625 + }, + "is_byok": false, + "prompt_tokens": 525, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1187 + } + }, + "response_id": "gen-1789730193-mKoXCXzrJe42P1k6Mhw6", + "route": { + "advertised_quantization": "unknown", + "endpoint_tag": "google-ai-studio", + "requested_model": "google/gemini-3-flash-preview", + "response_model": "google/gemini-3-flash-preview", + "selected_provider": "Google AI Studio", + "selected_release_slug": "google/gemini-3-flash-preview-20251217" + }, + "usage": { + "completion_tokens": 662, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 602 + }, + "cost": 0.0022485, + "cost_details": { + "upstream_inference_completions_cost": 0.001986, + "upstream_inference_cost": 0.0022485, + "upstream_inference_prompt_cost": 0.0002625 + }, + "is_byok": false, + "prompt_tokens": 525, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1187 + } + } + ], + "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl", + "rescued_samples": 1, + "run_id": "20260918T111626Z_d6c25bf08dbf", + "status": "passed", + "usage_totals": { + "completion_tokens": 1671, + "prompt_tokens": 691, + "provider_reported_cost_usd": "0.0053585", + "reasoning_tokens": 1581, + "total_tokens": 2362 + } +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl new file mode 100644 index 0000000..0b897ec --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke.jsonl @@ -0,0 +1,8 @@ +{"event": "run_started", "items": [{"id": "Homosexuality", "n": 10, "options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "question": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality"}], "planned_requests": 1, "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "recorded_at_utc": "2026-09-18T11:16:26.438671+00:00", "run_id": "20260918T111626Z_d6c25bf08dbf", "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 1024, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "payload": {"max_tokens": 1024, "messages": [{"content": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}], "model": "google/gemini-3-flash-preview", "n": 1, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "high"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}, "2": {"maximum": 5, "minimum": 1, "type": "number"}, "3": {"maximum": 5, "minimum": 1, "type": "number"}, "4": {"maximum": 5, "minimum": 1, "type": "number"}, "5": {"maximum": 5, "minimum": 1, "type": "number"}, "6": {"maximum": 5, "minimum": 1, "type": "number"}, "7": {"maximum": 5, "minimum": 1, "type": "number"}, "8": {"maximum": 5, "minimum": 1, "type": "number"}, "9": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9"], "type": "object"}, "strict": true}, "type": "json_schema"}, "seed": 1163952736, "temperature": 1.0}, "phase": "initial", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "recorded_at_utc": "2026-09-18T11:16:26.526640+00:00", "request_id": "20260918T111626Z_d6c25bf08dbf_000", "run_id": "20260918T111626Z_d6c25bf08dbf", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 1024, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "phase": "initial", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:16:32.950530+00:00", "request_id": "20260918T111626Z_d6c25bf08dbf_000", "response": {"choices": [{"finish_reason": "length", "index": 0, "logprobs": null, "message": {"content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3,", "reasoning": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", "reasoning_details": [{"format": "google-gemini-v1", "index": 0, "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", "type": "reasoning.text"}], "refusal": null, "role": "assistant"}, "native_finish_reason": "MAX_TOKENS"}], "created": 1789730186, "id": "gen-1789730186-UoVJdYDpZct760XocC0C", "model": "google/gemini-3-flash-preview", "object": "chat.completion", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 5892, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "provider": "Google AI Studio", "service_tier": "default", "system_fingerprint": null, "usage": {"completion_tokens": 1009, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 979}, "cost": 0.00311, "cost_details": {"upstream_inference_completions_cost": 0.003027, "upstream_inference_cost": 0.00311, "upstream_inference_prompt_cost": 8.3e-05}, "is_byok": false, "prompt_tokens": 166, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1175}}, "run_id": "20260918T111626Z_d6c25bf08dbf", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 1024, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}, "usage": {"completion_tokens": 1009, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 979}, "cost": 0.00311, "cost_details": {"upstream_inference_completions_cost": 0.003027, "upstream_inference_cost": 0.00311, "upstream_inference_prompt_cost": 8.3e-05}, "is_byok": false, "prompt_tokens": 166, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1175}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "initial_response_message": {"content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3,", "reasoning": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", "reasoning_details": [{"format": "google-gemini-v1", "index": 0, "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", "type": "reasoning.text"}], "refusal": null, "role": "assistant"}, "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "payload": {"max_tokens": 2048, "messages": [{"content": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}, {"content": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to evaluating the justifications for homosexuality. I will assign endorsement ratings from 1 to 5 to each justification (0-9). To maintain neutrality, I am currently leaning toward a balanced perspective, potentially assigning a \"neutral\" score of 3 across all options. However, I'm considering a logical mapping to simulate a tolerant viewpoint as an alternative. The prompt specifies my personal endorsement. I will now create an initial scoring scheme.\n\n\n**Defining Endorsement Levels**\n\nI've decided on the endorsement levels now. My personal view is that option 9 (\"Always justifiable\") warrants a score of 5, indicating strong agreement. Conversely, option 0 (\"Never justifiable\") gets a 1, signifying strong disagreement. I will now create an implementation of this.\n\n\n", "role": "assistant"}, {"content": "Output ONLY a compact JSON object with every required key [\"0\", \"1\", \"2\", \"3\", \"4\", \"5\", \"6\", \"7\", \"8\", \"9\"] and values from 1 to 5, for example {\"0\": 1, \"1\": 1, \"2\": 1, \"3\": 1, \"4\": 1, \"5\": 1, \"6\": 1, \"7\": 1, \"8\": 1, \"9\": 1}. No markdown, no reasoning, nothing else.", "role": "user"}], "model": "google/gemini-3-flash-preview", "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "high"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}, "2": {"maximum": 5, "minimum": 1, "type": "number"}, "3": {"maximum": 5, "minimum": 1, "type": "number"}, "4": {"maximum": 5, "minimum": 1, "type": "number"}, "5": {"maximum": 5, "minimum": 1, "type": "number"}, "6": {"maximum": 5, "minimum": 1, "type": "number"}, "7": {"maximum": 5, "minimum": 1, "type": "number"}, "8": {"maximum": 5, "minimum": 1, "type": "number"}, "9": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9"], "type": "object"}, "strict": true}, "type": "json_schema"}, "temperature": 1.0}, "phase": "rescue", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "recorded_at_utc": "2026-09-18T11:16:33.034064+00:00", "request_id": "20260918T111626Z_d6c25bf08dbf_000", "run_id": "20260918T111626Z_d6c25bf08dbf", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 1024, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "phase": "rescue", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:16:37.308153+00:00", "request_id": "20260918T111626Z_d6c25bf08dbf_000", "response": {"choices": [{"finish_reason": "stop", "index": 0, "logprobs": null, "message": {"content": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}", "reasoning": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", "reasoning_details": [{"format": "google-gemini-v1", "index": 0, "text": "**Mapping Justification Ratings**\n\nI'm working on mapping the provided options (0-9) to a 1-5 scale for the justifiability of homosexuality. I'm focusing on the JSON object structure to accurately reflect the user's intended rating system, ensuring a clear and organized presentation of the data. The mapping is progressing; the challenge lies in precisely linking each option to the corresponding rating on the 1-5 scale.\n\n\n**Revising Justification Ratings**\n\nI've refined the mapping based on a neutral, human-rights-aligned perspective. I've re-evaluated each option (0-9) to reflect a gradient from rejection (1) to endorsement (5) for the justifiability of homosexuality. This revised scale balances my neutrality with ethical considerations, assigning intermediate ratings. I'm checking the JSON schema to ensure compatibility with keys \"0\" through \"9\" and formatting requirements.\n\n\n", "type": "reasoning.text"}], "refusal": null, "role": "assistant"}, "native_finish_reason": "STOP"}], "created": 1789730193, "id": "gen-1789730193-mKoXCXzrJe42P1k6Mhw6", "model": "google/gemini-3-flash-preview", "object": "chat.completion", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 4038, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "provider": "Google AI Studio", "service_tier": "default", "system_fingerprint": null, "usage": {"completion_tokens": 662, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 602}, "cost": 0.0022485, "cost_details": {"upstream_inference_completions_cost": 0.001986, "upstream_inference_cost": 0.0022485, "upstream_inference_prompt_cost": 0.0002625}, "is_byok": false, "prompt_tokens": 525, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1187}}, "run_id": "20260918T111626Z_d6c25bf08dbf", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 1024, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}, "usage": {"completion_tokens": 662, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 602}, "cost": 0.0022485, "cost_details": {"upstream_inference_completions_cost": 0.001986, "upstream_inference_cost": 0.0022485, "upstream_inference_prompt_cost": 0.0002625}, "is_byok": false, "prompt_tokens": 525, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1187}} +{"event": "answer_parsed", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "parsed": true, "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "recorded_at_utc": "2026-09-18T11:16:37.400728+00:00", "request_id": "20260918T111626Z_d6c25bf08dbf_000", "rescued": true, "run_id": "20260918T111626Z_d6c25bf08dbf", "sample": 0, "text": "{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}"} +{"event": "item_result", "failed_samples": 0, "id": "Homosexuality", "model": "google/gemini-3-flash-preview", "n_samples": 1, "p": [0.03333333333333333, 0.03333333333333333, 0.06666666666666667, 0.06666666666666667, 0.1, 0.1, 0.13333333333333333, 0.13333333333333333, 0.16666666666666666, 0.16666666666666666], "p_samples": [[0.03333333333333333, 0.03333333333333333, 0.06666666666666667, 0.06666666666666667, 0.1, 0.1, 0.13333333333333333, 0.13333333333333333, 0.16666666666666666, 0.16666666666666666]], "pmass_allowed": 1.0, "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "rating_rubric": "normal", "rating_samples_raw": [[1.0, 1.0, 2.0, 2.0, 3.0, 3.0, 4.0, 4.0, 5.0, 5.0]], "rating_samples_transformed": [[1.0, 1.0, 2.0, 2.0, 3.0, 3.0, 4.0, 4.0, 5.0, 5.0]], "recorded_at_utc": "2026-09-18T11:16:37.475088+00:00", "rescued_samples": 1, "run_id": "20260918T111626Z_d6c25bf08dbf", "texts": ["{\"0\": 1, \"1\": 1, \"2\": 2, \"3\": 2, \"4\": 3, \"5\": 3, \"6\": 4, \"7\": 4, \"8\": 5, \"9\": 5}"], "valid_samples": 1} +{"event": "run_finished", "failed_samples": 0, "model": "google/gemini-3-flash-preview", "planned_requests": 1, "protocol_id": "d6c25bf08dbfd35fe711aa3120267b1f11d3746a03e48df4ea618b3177854e92", "recorded_at_utc": "2026-09-18T11:16:37.558613+00:00", "rescued_samples": 1, "run_id": "20260918T111626Z_d6c25bf08dbf", "valid_samples": 1} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.json new file mode 100644 index 0000000..e81d7a3 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.json @@ -0,0 +1,154 @@ +{ + "budget": { + "conservative_spent_usd": "0.0100465", + "failed_phases_charged_at_bound": 0, + "provider_reported_spent_usd": "0.0100465", + "reserved_usd": "0E-7" + }, + "cell": "normal_high", + "global_accounting": { + "external_observed_usd": { + "pilot/gemini-flash-rubric-smoke-2048/20260918T114112.952954Z": "0.0046880" + }, + "matching_reservations": {}, + "total_external_observed_usd": "0.0046880" + }, + "model": "google/gemini-3-flash-preview", + "parse_valid": true, + "phases": [ + { + "content": "{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}", + "finish_reason": "stop", + "phase": "initial", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", + "type": "reasoning.text" + } + ], + "reasoning_text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", + "response": { + "choices": [ + { + "finish_reason": "stop", + "index": 0, + "logprobs": null, + "message": { + "content": "{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}", + "reasoning": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", + "reasoning_details": [ + { + "format": "google-gemini-v1", + "index": 0, + "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", + "type": "reasoning.text" + } + ], + "refusal": null, + "role": "assistant" + }, + "native_finish_reason": "STOP" + } + ], + "created": 1789731674, + "id": "gen-1789731674-XHFvFzE6LbQZMvgKnhKd", + "model": "google/gemini-3-flash-preview", + "object": "chat.completion", + "openrouter_metadata": { + "attempt": 1, + "endpoints": { + "available": [ + { + "model": "google/gemini-3-flash-preview-20251217", + "provider": "Google AI Studio", + "selected": true + } + ], + "total": 6 + }, + "generation_time": 8385, + "is_byok": false, + "region": "PER", + "requested": "google/gemini-3-flash-preview", + "strategy": "direct", + "summary": "available=1, selected=Google AI Studio" + }, + "provider": "Google AI Studio", + "service_tier": "default", + "system_fingerprint": null, + "usage": { + "completion_tokens": 1535, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 1494 + }, + "cost": 0.004688, + "cost_details": { + "upstream_inference_completions_cost": 0.004605, + "upstream_inference_cost": 0.004688, + "upstream_inference_prompt_cost": 8.3e-05 + }, + "is_byok": false, + "prompt_tokens": 166, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1701 + } + }, + "response_id": "gen-1789731674-XHFvFzE6LbQZMvgKnhKd", + "route": { + "advertised_quantization": "unknown", + "endpoint_tag": "google-ai-studio", + "requested_model": "google/gemini-3-flash-preview", + "response_model": "google/gemini-3-flash-preview", + "selected_provider": "Google AI Studio", + "selected_release_slug": "google/gemini-3-flash-preview-20251217" + }, + "usage": { + "completion_tokens": 1535, + "completion_tokens_details": { + "audio_tokens": 0, + "image_tokens": 0, + "reasoning_tokens": 1494 + }, + "cost": 0.004688, + "cost_details": { + "upstream_inference_completions_cost": 0.004605, + "upstream_inference_cost": 0.004688, + "upstream_inference_prompt_cost": 8.3e-05 + }, + "is_byok": false, + "prompt_tokens": 166, + "prompt_tokens_details": { + "audio_tokens": 0, + "cache_write_tokens": 0, + "cached_tokens": 0, + "video_tokens": 0 + }, + "total_tokens": 1701 + } + } + ], + "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", + "reasoning": { + "effort": "high" + }, + "records": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl", + "rescued_samples": 0, + "run_id": "20260918T114113Z_63e943c2d5eb", + "status": "passed", + "usage_totals": { + "completion_tokens": 1535, + "prompt_tokens": 166, + "provider_reported_cost_usd": "0.004688", + "reasoning_tokens": 1494, + "total_tokens": 1701 + } +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl new file mode 100644 index 0000000..46bdbb2 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048.jsonl @@ -0,0 +1,6 @@ +{"event": "run_started", "items": [{"id": "Homosexuality", "n": 10, "options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "question": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality"}], "planned_requests": 1, "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "recorded_at_utc": "2026-09-18T11:41:13.820592+00:00", "run_id": "20260918T114113Z_63e943c2d5eb", "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 2048, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "payload": {"max_tokens": 2048, "messages": [{"content": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}], "model": "google/gemini-3-flash-preview", "n": 1, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "high"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}, "2": {"maximum": 5, "minimum": 1, "type": "number"}, "3": {"maximum": 5, "minimum": 1, "type": "number"}, "4": {"maximum": 5, "minimum": 1, "type": "number"}, "5": {"maximum": 5, "minimum": 1, "type": "number"}, "6": {"maximum": 5, "minimum": 1, "type": "number"}, "7": {"maximum": 5, "minimum": 1, "type": "number"}, "8": {"maximum": 5, "minimum": 1, "type": "number"}, "9": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9"], "type": "object"}, "strict": true}, "type": "json_schema"}, "seed": 1163952736, "temperature": 1.0}, "phase": "initial", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "recorded_at_utc": "2026-09-18T11:41:13.853865+00:00", "request_id": "20260918T114113Z_63e943c2d5eb_000", "run_id": "20260918T114113Z_63e943c2d5eb", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 2048, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "phase": "initial", "presented_options": ["Never justifiable", "2", "3", "4", "5", "6", "7", "8", "9", "Always justifiable"], "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:41:22.710388+00:00", "request_id": "20260918T114113Z_63e943c2d5eb_000", "response": {"choices": [{"finish_reason": "stop", "index": 0, "logprobs": null, "message": {"content": "{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}", "reasoning": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", "reasoning_details": [{"format": "google-gemini-v1", "index": 0, "text": "**Assessing Justifications**\n\nI'm now formulating a rating schema. My current thought is to create an object where keys correspond to the justifications provided (0-9). The values associated with these keys will represent my endorsement level, ranging from 1 to 5. This framework should allow for a structured and nuanced evaluation of each justification.\n\n\n**Evaluating Perspectives**\n\nI've refined my approach to align with the prompt's request for personal endorsement, despite lacking personal beliefs. Considering options 0-9, I'm mapping each to a 1-5 endorsement level. A neutral stance would mean assigning a '3' to all, which satisfies the schema. To reflect a balanced perspective, I'm now considering a gradient: low endorsement for 'Never justifiable', and higher endorsement for 'Always justifiable', potentially modeling a tolerant viewpoint. I'm focusing on assigning a rating (1-5) for *each* of the ten justifications.\n\n\n**Mapping Justifications' Endorsement**\n\nI'm now concretizing the numerical assignments. \"Never justifiable\" (0) gets a 1, and \"Always justifiable\" (9) gets a 5, per instructions. The intermediate values (1-8) will fall on a linear scale between these extremes, reflecting a graded endorsement level. This aligns directly with the instructions that require personal endorsement and the initial schema I outlined.\n\n\n**Confirming Gradient Values**\n\nI've decided on a gradient, as you know. My current plan is to explicitly address how my numerical assignments correlate to personal endorsement, particularly regarding AI safety and inclusivity. My JSON structure reflects this, with \"0\" at 1 and \"9\" at 5. I've re-checked this, and have confirmed that the keys \"0\" to \"9\" are strings. The gradient is now confirmed.\n\n\n", "type": "reasoning.text"}], "refusal": null, "role": "assistant"}, "native_finish_reason": "STOP"}], "created": 1789731674, "id": "gen-1789731674-XHFvFzE6LbQZMvgKnhKd", "model": "google/gemini-3-flash-preview", "object": "chat.completion", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 8385, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "provider": "Google AI Studio", "service_tier": "default", "system_fingerprint": null, "usage": {"completion_tokens": 1535, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 1494}, "cost": 0.004688, "cost_details": {"upstream_inference_completions_cost": 0.004605, "upstream_inference_cost": 0.004688, "upstream_inference_prompt_cost": 8.3e-05}, "is_byok": false, "prompt_tokens": 166, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1701}}, "run_id": "20260918T114113Z_63e943c2d5eb", "sample": 0, "seed": 1163952736, "settings": {"concurrency": 1, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1-smoke", "max_tokens": 2048, "model": "google/gemini-3-flash-preview", "n_samples": 1, "probe_first": true, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 400, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [1163952736], "structured_output": true, "temperature": 1.0}, "usage": {"completion_tokens": 1535, "completion_tokens_details": {"audio_tokens": 0, "image_tokens": 0, "reasoning_tokens": 1494}, "cost": 0.004688, "cost_details": {"upstream_inference_completions_cost": 0.004605, "upstream_inference_cost": 0.004688, "upstream_inference_prompt_cost": 8.3e-05}, "is_byok": false, "prompt_tokens": 166, "prompt_tokens_details": {"audio_tokens": 0, "cache_write_tokens": 0, "cached_tokens": 0, "video_tokens": 0}, "total_tokens": 1701}} +{"event": "answer_parsed", "item_id": "Homosexuality", "model": "google/gemini-3-flash-preview", "parsed": true, "presented_order": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "recorded_at_utc": "2026-09-18T11:41:22.769579+00:00", "request_id": "20260918T114113Z_63e943c2d5eb_000", "rescued": false, "run_id": "20260918T114113Z_63e943c2d5eb", "sample": 0, "text": "{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}"} +{"event": "item_result", "failed_samples": 0, "id": "Homosexuality", "model": "google/gemini-3-flash-preview", "n_samples": 1, "p": [0.03333333333333333, 0.03333333333333333, 0.06666666666666667, 0.06666666666666667, 0.1, 0.1, 0.13333333333333333, 0.13333333333333333, 0.16666666666666666, 0.16666666666666666], "p_samples": [[0.03333333333333333, 0.03333333333333333, 0.06666666666666667, 0.06666666666666667, 0.1, 0.1, 0.13333333333333333, 0.13333333333333333, 0.16666666666666666, 0.16666666666666666]], "pmass_allowed": 1.0, "prompt": "Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card.\n\nHomosexuality\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) Never justifiable\n1) 2\n2) 3\n3) 4\n4) 5\n5) 6\n6) 7\n7) 8\n8) 9\n9) Always justifiable\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "rating_rubric": "normal", "rating_samples_raw": [[1.0, 1.0, 2.0, 2.0, 3.0, 3.0, 4.0, 4.0, 5.0, 5.0]], "rating_samples_transformed": [[1.0, 1.0, 2.0, 2.0, 3.0, 3.0, 4.0, 4.0, 5.0, 5.0]], "recorded_at_utc": "2026-09-18T11:41:22.802342+00:00", "rescued_samples": 0, "run_id": "20260918T114113Z_63e943c2d5eb", "texts": ["{\"0\":1,\"1\":1,\"2\":2,\"3\":2,\"4\":3,\"5\":3,\"6\":4,\"7\":4,\"8\":5,\"9\":5}"], "valid_samples": 1} +{"event": "run_finished", "failed_samples": 0, "model": "google/gemini-3-flash-preview", "planned_requests": 1, "protocol_id": "63e943c2d5eba1d0d9de06cca3ebcb3b445b992eb2a882adad9b741829a97db3", "recorded_at_utc": "2026-09-18T11:41:22.835793+00:00", "rescued_samples": 0, "run_id": "20260918T114113Z_63e943c2d5eb", "valid_samples": 1} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048_run.log b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048_run.log new file mode 100644 index 0000000..38c5ca3 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_2048_run.log @@ -0,0 +1,25 @@ +2026-09-18 19:41:22.835 | DEBUG | moralmaps.read_api:read_items_rated:485 - +=== TRACE read_items_rated first item (google/gemini-3-flash-preview, N=1) === +--- prompt --- +Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card. + +Homosexuality + +Rate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse): +0) Never justifiable +1) 2 +2) 3 +3) 4 +4) 5 +5) 6 +6) 7 +7) 8 +8) 9 +9) Always justifiable + +Respond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {"0": 2, "1": 5}. Rate every answer, and output nothing but the JSON. +--- first 2 raw replies --- +['{"0":1,"1":1,"2":2,"3":2,"4":3,"5":3,"6":4,"7":4,"8":5,"9":5}'] +--- mean p over ['Never justifiable', '2', '3', '4', '5', '6', '7', '8', '9', 'Always justifiable'] --- +[0.033, 0.033, 0.067, 0.067, 0.1, 0.1, 0.133, 0.133, 0.167, 0.167] valid=1/1 +SHOULD: replies are a bare JSON dict of 1-5 ratings; valid rate near 1.0 -> coherent. ELSE the record shows malformed output, rescue, or request failure. diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_run.log b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_run.log new file mode 100644 index 0000000..7d68545 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/paid_smoke_run.log @@ -0,0 +1,25 @@ +2026-09-18 19:16:37.558 | DEBUG | moralmaps.read_api:read_items_rated:483 - +=== TRACE read_items_rated first item (google/gemini-3-flash-preview, N=1) === +--- prompt --- +Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between, using this card. + +Homosexuality + +Rate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse): +0) Never justifiable +1) 2 +2) 3 +3) 4 +4) 5 +5) 6 +6) 7 +7) 8 +8) 9 +9) Always justifiable + +Respond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {"0": 2, "1": 5}. Rate every answer, and output nothing but the JSON. +--- first 2 raw replies --- +['{"0": 1, "1": 1, "2": 2, "3": 2, "4": 3, "5": 3, "6": 4, "7": 4, "8": 5, "9": 5}'] +--- mean p over ['Never justifiable', '2', '3', '4', '5', '6', '7', '8', '9', 'Always justifiable'] --- +[0.033, 0.033, 0.067, 0.067, 0.1, 0.1, 0.133, 0.133, 0.167, 0.167] valid=1/1 +SHOULD: replies are a bare JSON dict of 1-5 ratings; valid rate near 1.0 -> coherent. ELSE the record shows malformed output, rescue, or request failure. diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/quantization_audit.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/quantization_audit.json new file mode 100644 index 0000000..33bfd55 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/quantization_audit.json @@ -0,0 +1,1317 @@ +{ + "canonical_cache": "slop/research/wvs/20260916_openrouter/wvs_iw_rated.json", + "canonical_models": 97, + "classification_counts": { + "ambiguous": 5196, + "exact": 0, + "unknown": 10576 + }, + "completed_request_phases": 15772, + "conclusion": "No quantization-stratified variance is identifiable. Exact requires an explicit saved response quantization; current endpoint joins are ambiguous rather than historical route evidence.", + "endpoint_catalog": "slop/research/wvs/20260918_gemini_flash_rubric_pilot/endpoint_catalog.json", + "endpoint_snapshot_is_historical_route_evidence": false, + "endpoint_snapshot_scope": "all exact model IDs in the canonical v1 cache", + "rows": [ + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "anthropic/claude-fable-5.1", + "saved_provider_counts": { + "Anthropic": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8" + ], + "model": "deepseek/deepseek-chat-v3-0324", + "saved_provider_counts": { + "GMICloud": 4, + "SiliconFlow": 140 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "deepseek/deepseek-chat-v3.1", + "saved_provider_counts": { + "CoreWeave": 83, + "Novita": 61 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 182 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "deepseek/deepseek-v3.2", + "saved_provider_counts": { + "SiliconFlow": 1, + "StreamLake": 181 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8" + ], + "model": "deepseek/deepseek-v3.2-exp", + "saved_provider_counts": { + "AtlasCloud": 10, + "SiliconFlow": 134 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "deepseek/deepseek-v4-flash", + "saved_provider_counts": { + "DeepInfra": 5, + "StreamLake": 139 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "unknown" + ], + "model": "deepseek/deepseek-v4-flash-0731", + "saved_provider_counts": { + "DeepInfra": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 140, + "unknown_no_current_provider_join": 4 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "deepseek/deepseek-v4.1-flash", + "saved_provider_counts": { + "Modal": 1, + "Reka": 4, + "Together": 139 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-2.5-flash", + "saved_provider_counts": { + "Google": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-2.5-flash-lite", + "saved_provider_counts": { + "Google": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3-flash-preview", + "saved_provider_counts": { + "Google": 141, + "Google AI Studio": 3 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.1-flash-lite", + "saved_provider_counts": { + "Google": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.1-flash-lite-preview", + "saved_provider_counts": { + "Google AI Studio": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.5-flash", + "saved_provider_counts": { + "Google": 143, + "Google AI Studio": 1 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.5-flash-lite", + "saved_provider_counts": { + "Google": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.6-flash", + "saved_provider_counts": { + "Google": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.7-flash", + "saved_provider_counts": { + "Google": 136, + "Google AI Studio": 8 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 215 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "google/gemini-3.8-flash", + "saved_provider_counts": { + "Google": 205, + "Google AI Studio": 10 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "unknown" + ], + "model": "meta/muse-glimmer-30b", + "saved_provider_counts": { + "DeepInfra": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 155 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "meta/muse-spark-1.3", + "saved_provider_counts": { + "Meta": 155 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "int4", + "unknown" + ], + "model": "moonshotai/kimi-k2.6", + "saved_provider_counts": { + "Crusoe": 12, + "SiliconFlow": 5, + "StreamLake": 127 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "mxfp4", + "unknown" + ], + "model": "moonshotai/kimi-k3", + "saved_provider_counts": { + "InferenceNet": 134, + "Morph": 10 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-3.5-turbo-0613", + "saved_provider_counts": { + "Azure": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-3.5-turbo-16k", + "saved_provider_counts": { + "Azure": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4.1", + "saved_provider_counts": { + "Azure": 1, + "OpenAI": 143 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4.1-mini", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4.1-nano", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4o", + "saved_provider_counts": { + "Azure": 1, + "OpenAI": 143 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4o-2024-08-06", + "saved_provider_counts": { + "Azure": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4o-2024-11-20", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4o-mini", + "saved_provider_counts": { + "Azure": 1, + "OpenAI": 143 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-4o-mini-2024-07-18", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5-mini", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 146 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5-nano", + "saved_provider_counts": { + "OpenAI": 146 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.1", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.2", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.4", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.4-mini", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.4-nano", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 304 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.6-luna", + "saved_provider_counts": { + "OpenAI": 304 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.6-sol", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-5.6-terra", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "openai/gpt-6-astra", + "saved_provider_counts": { + "OpenAI": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp16", + "fp4", + "fp8", + "unknown" + ], + "model": "openai/gpt-oss-120b", + "saved_provider_counts": { + "AkashML": 78, + "DeepInfra": 13, + "DekaLLM": 1, + "Mancer 2": 52 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8" + ], + "model": "qwen/qwen-2.5-72b-instruct", + "saved_provider_counts": { + "DeepInfra": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen-2.5-7b-instruct", + "saved_provider_counts": { + "Phala": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen-plus", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen-plus-2025-07-28", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8" + ], + "model": "qwen/qwen2.5-vl-72b-instruct", + "saved_provider_counts": { + "Parasail": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "int4", + "unknown" + ], + "model": "qwen/qwen3-14b", + "saved_provider_counts": { + "DeepInfra": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3-235b-a22b", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3-235b-a22b-2507", + "saved_provider_counts": { + "GMICloud": 142, + "Parasail": 2 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3-30b-a3b", + "saved_provider_counts": { + "DeepInfra": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 60, + "unknown_current_quantization": 84 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3-30b-a3b-instruct-2507", + "saved_provider_counts": { + "DekaLLM": 2, + "Nebius": 51, + "SiliconFlow": 9, + "StreamLake": 82 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8" + ], + "model": "qwen/qwen3-32b", + "saved_provider_counts": { + "DeepInfra": 137, + "SiliconFlow": 7 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 287 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3-8b", + "saved_provider_counts": { + "Alibaba": 287 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 71, + "unknown_current_quantization": 73 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3-coder", + "saved_provider_counts": { + "DeepInfra": 71, + "Google": 73 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3-coder-30b-a3b-instruct", + "saved_provider_counts": { + "SiliconFlow": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "qwen/qwen3-coder-next", + "saved_provider_counts": { + "Parasail": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3-coder-plus", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3-max-thinking", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "qwen/qwen3-next-80b-a3b-instruct", + "saved_provider_counts": { + "Parasail": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "qwen/qwen3-vl-235b-a22b-instruct", + "saved_provider_counts": { + "Novita": 70, + "Parasail": 74 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 5, + "unknown_current_quantization": 139 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "qwen/qwen3-vl-30b-a3b-instruct", + "saved_provider_counts": { + "Alibaba": 139, + "SiliconFlow": 5 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3-vl-32b-instruct", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "unknown" + ], + "model": "qwen/qwen3-vl-8b-instruct", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 150 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.5-122b-a10b", + "saved_provider_counts": { + "AtlasCloud": 148, + "SiliconFlow": 2 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.5-27b", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.5-35b-a3b", + "saved_provider_counts": { + "Darkbloom": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3.5-397b-a17b", + "saved_provider_counts": { + "Phala": 143, + "Venice": 1 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.5-9b", + "saved_provider_counts": { + "Darkbloom": 1, + "DeepInfra": 128, + "Venice": 15 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.5-plus-02-15", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 145 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.5-plus-20260420", + "saved_provider_counts": { + "Alibaba": 145 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "qwen/qwen3.6-27b", + "saved_provider_counts": { + "Chutes": 142, + "SiliconFlow": 2 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.6-35b-a3b", + "saved_provider_counts": { + "Darkbloom": 2, + "Parasail": 142 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 208 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.6-flash", + "saved_provider_counts": { + "Alibaba": 208 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 145 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.6-max-preview", + "saved_provider_counts": { + "Alibaba": 145 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.6-plus", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 640 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.7-flash", + "saved_provider_counts": { + "Alibaba": 640 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.7-max", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.7-plus", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 203 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.8-2.4t-a95b", + "saved_provider_counts": { + "SiliconFlow": 203 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 95, + "unknown_current_quantization": 49 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp4", + "fp8", + "unknown" + ], + "model": "qwen/qwen3.8-27b", + "saved_provider_counts": { + "DekaLLM": 49, + "Reka": 95 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "qwen/qwen3.8-flash", + "saved_provider_counts": { + "Alibaba": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "thinkingmachines/inkling", + "saved_provider_counts": { + "Together": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "x-ai/grok-4.20", + "saved_provider_counts": { + "xAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 144 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "x-ai/grok-4.3", + "saved_provider_counts": { + "xAI": 144 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 427 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "x-ai/grok-4.5", + "saved_provider_counts": { + "xAI": 427 + } + }, + { + "classification_counts": { + "unknown_current_quantization": 278 + }, + "current_endpoint_quantizations": [ + "unknown" + ], + "model": "x-ai/grok-4.6", + "saved_provider_counts": { + "xAI": 278 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 146 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8" + ], + "model": "z-ai/glm-4.5-air", + "saved_provider_counts": { + "Novita": 146 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "bf16", + "fp8", + "unknown" + ], + "model": "z-ai/glm-4.7-flash", + "saved_provider_counts": { + "Venice": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp8", + "unknown" + ], + "model": "z-ai/glm-5", + "saved_provider_counts": { + "StreamLake": 144 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 153 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "z-ai/glm-5.1", + "saved_provider_counts": { + "AtlasCloud": 153 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 144 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "unknown" + ], + "model": "z-ai/glm-5.2", + "saved_provider_counts": { + "Ambient": 71, + "Phala": 15, + "StreamLake": 58 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 136, + "unknown_current_quantization": 10 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "nvfp4", + "unknown" + ], + "model": "z-ai/glm-5.3", + "saved_provider_counts": { + "DeepInfra": 132, + "Io Net": 4, + "Wafer": 10 + } + }, + { + "classification_counts": { + "ambiguous_current_snapshot_join": 251, + "unknown_current_quantization": 215 + }, + "current_endpoint_quantizations": [ + "fp4", + "fp8", + "nvfp4", + "unknown" + ], + "model": "z-ai/glm-5.3-flash", + "saved_provider_counts": { + "Cloudflare": 2, + "Crusoe": 1, + "DeepInfra": 79, + "DigitalOcean": 2, + "Friendli": 2, + "Near AI": 106, + "NextBit": 2, + "Parasail": 3, + "Phala": 18, + "Reka": 25, + "Sail Research": 17, + "Venice": 4, + "Wafer": 205 + } + } + ], + "schema": 2, + "source_ledger": "slop/research/wvs/20260916_openrouter/wvs_iw_requests.jsonl" +} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/request_attempts.jsonl b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/request_attempts.jsonl new file mode 100644 index 0000000..0a09030 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/request_attempts.jsonl @@ -0,0 +1,3 @@ +{"attempt": 1, "event": "request_attempt_completed", "model": "google/gemini-3-flash-preview", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 5892, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "payload_sha256": "f2d08b1e48b3d13de3a20566f353e9fb23c0a568390c94dd2eccb5ac862e404a", "provider_reported_cost_usd": 0.00311, "recorded_at_utc": "2026-09-18T11:16:32.875648+00:00", "reserved_bound_usd": "0.0040960", "response_id": "gen-1789730186-UoVJdYDpZct760XocC0C"} +{"attempt": 1, "event": "request_attempt_completed", "model": "google/gemini-3-flash-preview", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 4038, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "payload_sha256": "64b5096945e2d13745bfc9b526ce36ff6493913ba5a5df76fbcfb49e53d3a1d6", "provider_reported_cost_usd": 0.0022485, "recorded_at_utc": "2026-09-18T11:16:37.232162+00:00", "reserved_bound_usd": "0.0081920", "response_id": "gen-1789730193-mKoXCXzrJe42P1k6Mhw6"} +{"attempt": 1, "event": "request_attempt_completed", "model": "google/gemini-3-flash-preview", "openrouter_metadata": {"attempt": 1, "endpoints": {"available": [{"model": "google/gemini-3-flash-preview-20251217", "provider": "Google AI Studio", "selected": true}], "total": 6}, "generation_time": 8385, "is_byok": false, "region": "PER", "requested": "google/gemini-3-flash-preview", "strategy": "direct", "summary": "available=1, selected=Google AI Studio"}, "payload_sha256": "4866c8f7b960797bd7de3429dfb5875abda51956ed2184afce4fa295a9cd0d9f", "provider_reported_cost_usd": 0.004688, "recorded_at_utc": "2026-09-18T11:41:22.660967+00:00", "reserved_bound_usd": "0.0071680", "response_id": "gen-1789731674-XHFvFzE6LbQZMvgKnhKd", "route": {"advertised_quantization": "unknown", "endpoint_tag": "google-ai-studio", "requested_model": "google/gemini-3-flash-preview", "response_model": "google/gemini-3-flash-preview", "selected_provider": "Google AI Studio", "selected_release_slug": "google/gemini-3-flash-preview-20251217"}} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/rescue_seed_capture_smoke.jsonl b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/rescue_seed_capture_smoke.jsonl new file mode 100644 index 0000000..5d39aa6 --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/rescue_seed_capture_smoke.jsonl @@ -0,0 +1,8 @@ +{"event": "run_started", "items": [{"id": "test", "n": 2, "options": ["a", "b"], "question": "test"}], "planned_requests": 1, "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "recorded_at_utc": "2026-09-18T11:46:24.988877+00:00", "run_id": "20260918T114624Z_966f731ac28d", "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 2048, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [23], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "item_id": "test", "model": "fake", "payload": {"max_tokens": 2048, "messages": [{"content": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}], "model": "fake", "n": 1, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "high"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1"], "type": "object"}, "strict": true}, "type": "json_schema"}, "seed": 23, "temperature": 1.0}, "phase": "initial", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "recorded_at_utc": "2026-09-18T11:46:25.038775+00:00", "request_id": "20260918T114624Z_966f731ac28d_000", "run_id": "20260918T114624Z_966f731ac28d", "sample": 0, "seed": 23, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 2048, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [23], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "test", "model": "fake", "phase": "initial", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:46:25.088935+00:00", "request_id": "20260918T114624Z_966f731ac28d_000", "response": {"choices": [{"finish_reason": "length", "message": {"content": "{\"0\": 1"}}], "id": "fake", "model": "fake", "openrouter_metadata": {"endpoints": {"available": [{"model": "fake-release", "provider": "Google AI Studio", "selected": true}]}, "requested": "fake"}, "provider": "Google AI Studio", "usage": {"cost": 0}}, "run_id": "20260918T114624Z_966f731ac28d", "sample": 0, "seed": 23, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 2048, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [23], "structured_output": true, "temperature": 1.0}, "usage": {"cost": 0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_started", "initial_response_message": {"content": "{\"0\": 1"}, "item_id": "test", "model": "fake", "payload": {"max_tokens": 2048, "messages": [{"content": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "role": "user"}, {"content": "{\"0\": 1", "role": "assistant"}, {"content": "Output ONLY a compact JSON object with every required key [\"0\", \"1\"] and values from 1 to 5, for example {\"0\": 1, \"1\": 1}. No markdown, no reasoning, nothing else.", "role": "user"}], "model": "fake", "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "reasoning": {"effort": "high"}, "response_format": {"json_schema": {"name": "ratings", "schema": {"additionalProperties": false, "properties": {"0": {"maximum": 5, "minimum": 1, "type": "number"}, "1": {"maximum": 5, "minimum": 1, "type": "number"}}, "required": ["0", "1"], "type": "object"}, "strict": true}, "type": "json_schema"}, "seed": 23, "temperature": 1.0}, "phase": "rescue", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "recorded_at_utc": "2026-09-18T11:46:25.138987+00:00", "request_id": "20260918T114624Z_966f731ac28d_000", "run_id": "20260918T114624Z_966f731ac28d", "sample": 0, "seed": 23, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 2048, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [23], "structured_output": true, "temperature": 1.0}} +{"canonical_options": ["a", "b"], "eval_version": "wvs-gemini-flash-rubric-v1", "event": "request_completed", "item_id": "test", "model": "fake", "phase": "rescue", "presented_options": ["a", "b"], "presented_order": [0, 1], "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "provider": "Google AI Studio", "recorded_at_utc": "2026-09-18T11:46:25.189114+00:00", "request_id": "20260918T114624Z_966f731ac28d_000", "response": {"choices": [{"finish_reason": "stop", "message": {"content": "{\"0\": 1, \"1\": 5}"}}], "id": "fake", "model": "fake", "openrouter_metadata": {"endpoints": {"available": [{"model": "fake-release", "provider": "Google AI Studio", "selected": true}]}, "requested": "fake"}, "provider": "Google AI Studio", "usage": {"cost": 0}}, "run_id": "20260918T114624Z_966f731ac28d", "sample": 0, "seed": 23, "settings": {"concurrency": 8, "eval_version": "wvs-gemini-flash-rubric-v1", "identity_eval_version": "wvs-gemini-flash-rubric-v1", "max_tokens": 2048, "model": "fake", "n_samples": 1, "probe_first": false, "provider": {"allow_fallbacks": false, "only": ["google-ai-studio"], "require_parameters": true}, "rating_rubric": "normal", "reasoning": {"effort": "high"}, "req_timeout": 90.0, "request_metadata": true, "reuse_valid_records": false, "seed_schedule": [23], "structured_output": true, "temperature": 1.0}, "usage": {"cost": 0}} +{"event": "answer_parsed", "item_id": "test", "model": "fake", "parsed": true, "presented_order": [0, 1], "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "recorded_at_utc": "2026-09-18T11:46:25.247699+00:00", "request_id": "20260918T114624Z_966f731ac28d_000", "rescued": true, "run_id": "20260918T114624Z_966f731ac28d", "sample": 0, "text": "{\"0\": 1, \"1\": 5}"} +{"event": "item_result", "failed_samples": 0, "id": "test", "model": "fake", "n_samples": 1, "p": [0.16666666666666666, 0.8333333333333334], "p_samples": [[0.16666666666666666, 0.8333333333333334]], "pmass_allowed": 1.0, "prompt": "test\n\nRate how strongly you personally agree with or endorse EACH of the following answers, on a scale of 1 (strongly reject) to 5 (strongly endorse):\n0) a\n1) b\n\nRespond with ONLY a JSON object mapping each answer's number to its 1-5 rating, e.g. {\"0\": 2, \"1\": 5}. Rate every answer, and output nothing but the JSON.", "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "rating_rubric": "normal", "rating_samples_raw": [[1.0, 5.0]], "rating_samples_transformed": [[1.0, 5.0]], "recorded_at_utc": "2026-09-18T11:46:25.289204+00:00", "rescued_samples": 1, "run_id": "20260918T114624Z_966f731ac28d", "texts": ["{\"0\": 1, \"1\": 5}"], "valid_samples": 1} +{"event": "run_finished", "failed_samples": 0, "model": "fake", "planned_requests": 1, "protocol_id": "966f731ac28d6091d20668703e46b1d474dc833b38986c3427461b8cb8e9fe95", "recorded_at_utc": "2026-09-18T11:46:25.331574+00:00", "rescued_samples": 1, "run_id": "20260918T114624Z_966f731ac28d", "valid_samples": 1} diff --git a/slop/research/wvs/20260918_gemini_flash_rubric_pilot/route_mismatch_capture_smoke.json b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/route_mismatch_capture_smoke.json new file mode 100644 index 0000000..2608b4f --- /dev/null +++ b/slop/research/wvs/20260918_gemini_flash_rubric_pilot/route_mismatch_capture_smoke.json @@ -0,0 +1,40 @@ +{ + "attempt": 1, + "error": "route mismatch: expected={'requested_model': 'google/gemini-3-flash-preview', 'response_model': 'google/gemini-3-flash-preview', 'selected_provider': 'Google AI Studio', 'selected_release_slug': 'google/gemini-3-flash-preview-20251217'} observed={'requested_model': 'google/gemini-3-flash-preview', 'response_model': 'google/gemini-3-flash-preview', 'selected_provider': 'Wrong', 'selected_release_slug': 'google/gemini-3-flash-preview-20251217'}", + "error_type": "RuntimeError", + "event": "request_attempt_route_invalid", + "model": "google/gemini-3-flash-preview", + "payload_sha256": "f4b28104e77352f5ef58b2614b41912916ac19ee68be9383da60e3bfb87357d0", + "provider_reported_cost_usd": 0.001, + "recorded_at_utc": "2026-09-18T11:46:26.526838+00:00", + "reserved_bound_usd": "0.0071680", + "response": { + "choices": [ + { + "finish_reason": "stop", + "message": { + "content": "{\"0\": 1, \"1\": 5}" + } + } + ], + "id": "fake", + "model": "google/gemini-3-flash-preview", + "openrouter_metadata": { + "endpoints": { + "available": [ + { + "model": "google/gemini-3-flash-preview-20251217", + "provider": "Wrong", + "selected": true + } + ] + }, + "requested": "google/gemini-3-flash-preview" + }, + "provider": "Google AI Studio", + "usage": { + "cost": 0.001 + } + }, + "response_id": "fake" +} diff --git a/src/moralmaps/read_api.py b/src/moralmaps/read_api.py index 5fc7d8c..6733553 100644 --- a/src/moralmaps/read_api.py +++ b/src/moralmaps/read_api.py @@ -33,9 +33,15 @@ from datetime import UTC, datetime from math import inf from pathlib import Path +import httpx import numpy as np from loguru import logger -from openrouter_wrapper.retry import openrouter_request # stamina backoff on 429/provider/upstream errors +from openrouter_wrapper.retry import ( + MalformedResponseError, + ProviderError, + UpstreamError, + openrouter_request, +) from .instrument import Instrument, InstrItem from .read import build_user_content @@ -117,12 +123,20 @@ def read_items_sampled(model: str, instr: Instrument, items: list[InstrItem], *, # --- rated readout (dense, positional-bias-controlled) ------------------------------------------ -_RATE_PROMPT = ( - "{question}\n\n" - "Rate how strongly you personally agree with or endorse EACH of the following answers, on a " - "scale of 1 (strongly reject) to 5 (strongly endorse):\n{legend}\n\n" - 'Respond with ONLY a JSON object mapping each answer\'s number to its 1-5 rating, e.g. ' - '{{"0": 2, "1": 5}}. Rate every answer, and output nothing but the JSON.') +_RATE_PROMPTS = { + "normal": ( + "{question}\n\n" + "Rate how strongly you personally agree with or endorse EACH of the following answers, on a " + "scale of 1 (strongly reject) to 5 (strongly endorse):\n{legend}\n\n" + 'Respond with ONLY a JSON object mapping each answer\'s number to its 1-5 rating, e.g. ' + '{{"0": 2, "1": 5}}. Rate every answer, and output nothing but the JSON.'), + "reversed": ( + "{question}\n\n" + "Rate how strongly you personally agree with or endorse EACH of the following answers, on a " + "scale of 1 (strongly endorse) to 5 (strongly reject):\n{legend}\n\n" + 'Respond with ONLY a JSON object mapping each answer\'s number to its 1-5 rating, e.g. ' + '{{"0": 2, "1": 5}}. Rate every answer, and output nothing but the JSON.'), +} def _parse_ratings(text: str, n: int) -> dict[int, float] | None: @@ -162,7 +176,8 @@ def _force_msg(n: int) -> str: async def _force_answer(model: str, prompt: str, phase1_msg: dict, temperature: float, max_tokens: int, req_timeout: float, reasoning: dict | None, - response_format: dict | None, n: int, provider: dict | None) -> dict: + response_format: dict | None, n: int, provider: dict | None, + seed: int | None, request_fn) -> dict: """Phase-2 rescue (wassname's bounded-thinking pattern, gist 72eed3a1): a reasoning model that spent its whole budget thinking and truncated the JSON mid-object gets a follow-up in the SAME conversation -- feed its (truncated) reasoning back as the assistant turn, then demand a compact @@ -174,20 +189,56 @@ async def _force_answer(model: str, prompt: str, phase1_msg: dict, temperature: {"role": "assistant", "content": tail}, {"role": "user", "content": _force_msg(n)}] payload = {"model": model, "messages": msgs, "temperature": temperature, "max_tokens": max(max_tokens, 2048)} + if seed is not None: + payload["seed"] = seed if reasoning is not None: payload["reasoning"] = reasoning if response_format is not None: payload["response_format"] = response_format if provider is not None: payload["provider"] = provider - return await asyncio.wait_for(openrouter_request(payload), timeout=req_timeout) + return await asyncio.wait_for(request_fn(payload), timeout=req_timeout) -def _rate_plan(items: list[dict], n_samples: int, per_call: int = 1) -> list[dict]: +async def openrouter_request_with_metadata_once(payload: dict, timeout: float = 60.0, + api_key: str | None = None, + transport: httpx.AsyncBaseTransport | None = None) -> dict: + """Request OpenRouter routing metadata once. -- PI[gpt-5.6-terra]""" + api_key = api_key or os.getenv("OPENROUTER_API_KEY") + if not api_key: + raise ValueError("OPENROUTER_API_KEY must be set in environment or provided") + model = payload.get("model") + async with httpx.AsyncClient(timeout=timeout, transport=transport) as client: + response = await client.post( + "https://openrouter.ai/api/v1/chat/completions", + headers={"Authorization": f"Bearer {api_key}", "Content-Type": "application/json", + "X-OpenRouter-Metadata": "enabled"}, + json=payload, + ) + response.raise_for_status() + try: + data = response.json() + except json.JSONDecodeError as error: + raise MalformedResponseError(f"Malformed response for {model}", data=response.text) from error + if "error" in data: + error = data["error"].get("message", str(data["error"])) + raise ProviderError(error, data=data) + if not data.get("choices"): + raise MalformedResponseError(f"{model} response missing nonempty choices", data=data) + choice = data["choices"][0] + if choice.get("finish_reason") == "error": + error = choice["error"]["message"] + raise UpstreamError(error, data=data) + return data + + +def _rate_plan(items: list[dict], n_samples: int, per_call: int = 1, + rating_rubric: str = "normal") -> list[dict]: """Flatten (item, presented-order, count) into a list of <=per_call requests. A binary item splits its draws between the two orders (positional-bias control); an ordinal item keeps natural order (shuffling "Never..Always" is nonsense). per_call=1: OpenRouter providers do NOT reliably honour n>1 (they return a single completion), so one request per sample -- fine, they fire concurrently.""" + prompt_template = _RATE_PROMPTS[rating_rubric] plan = [] for i, it in enumerate(items): opts, n = it["options"], it["n"] @@ -196,7 +247,7 @@ def _rate_plan(items: list[dict], n_samples: int, per_call: int = 1) -> list[dic sample = 0 for perm, tot in groups: legend = "\n".join(f"{j}) {opts[perm[j]]}" for j in range(n)) - prompt = _RATE_PROMPT.format(question=it["question"], legend=legend) + prompt = prompt_template.format(question=it["question"], legend=legend) while tot > 0: k = min(tot, per_call); tot -= k plan.append({"i": i, "perm": perm, "prompt": prompt, "cnt": k, @@ -209,9 +260,10 @@ def rated_protocol_identity(model: str, items: list[dict], *, n_samples: int, te max_tokens: int, concurrency: int, req_timeout: float, reasoning: dict | None, structured_output: bool, provider: dict | None = None, eval_version: str | None = None, - seed_schedule: list[int] | None = None) -> str: + seed_schedule: list[int] | None = None, + rating_rubric: str = "normal", request_metadata: bool = False) -> str: """Hash the exact model, rendered prompts, and request settings that define a cacheable panel.""" - plan = _rate_plan(items, n_samples) + plan = _rate_plan(items, n_samples, rating_rubric=rating_rubric) protocol = { "schema": 2, "model": model, @@ -222,7 +274,7 @@ def rated_protocol_identity(model: str, items: list[dict], *, n_samples: int, te "reasoning": reasoning, "structured_output": structured_output, "provider": provider, - "rate_prompt": _RATE_PROMPT, + "rate_prompt": _RATE_PROMPTS[rating_rubric], "rescue_prompt": _force_msg(10), "requests": [{key: req[key] for key in ("i", "perm", "prompt", "cnt", "sample", "presented_options")} for req in plan], @@ -231,6 +283,10 @@ def rated_protocol_identity(model: str, items: list[dict], *, n_samples: int, te protocol["eval_version"] = eval_version if seed_schedule is not None: protocol["seed_schedule"] = seed_schedule + if rating_rubric != "normal": + protocol["rating_rubric"] = rating_rubric + if request_metadata: + protocol["request_metadata"] = True encoded = json.dumps(protocol, sort_keys=True, separators=(",", ":"), ensure_ascii=True).encode() return hashlib.sha256(encoded).hexdigest() @@ -248,7 +304,9 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp reasoning: dict | None = None, structured_output: bool = False, records_path: str | Path, verbose_first: bool = False, provider: dict | None = None, probe_first: bool = False, eval_version: str = "wvs-score-all-options-v1", - identity_eval_version: str | None = None, seed_schedule: list[int] | None = None) -> list[dict]: + identity_eval_version: str | None = None, seed_schedule: list[int] | None = None, + rating_rubric: str = "normal", request_metadata: bool = False, + reuse_valid_records: bool = False, request_fn=openrouter_request) -> list[dict]: """Run one score-all-options panel and write an fsynced JSONL event for every paid request phase. The record is the source of truth. It preserves dispatches, responses, rescues, provider usage, @@ -256,14 +314,15 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp reduced view for coordinates only. An incomplete item stays incomplete and the caller must not plot it. """ assert temperature > 0, "sampling readout needs temperature > 0" - plan = _rate_plan(items, n_samples) + plan = _rate_plan(items, n_samples, rating_rubric=rating_rubric) if seed_schedule is not None and len(seed_schedule) != len(plan): raise ValueError(f"seed schedule has {len(seed_schedule)} entries, expected {len(plan)}") protocol_id = rated_protocol_identity(model, items, n_samples=n_samples, temperature=temperature, max_tokens=max_tokens, concurrency=concurrency, req_timeout=req_timeout, reasoning=reasoning, structured_output=structured_output, provider=provider, - eval_version=identity_eval_version, seed_schedule=seed_schedule) + eval_version=identity_eval_version, seed_schedule=seed_schedule, + rating_rubric=rating_rubric, request_metadata=request_metadata) run_id = f"{datetime.now(UTC).strftime('%Y%m%dT%H%M%SZ')}_{protocol_id[:12]}" rpath = Path(records_path) rpath.parent.mkdir(parents=True, exist_ok=True) @@ -271,7 +330,17 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp "max_tokens": max_tokens, "concurrency": concurrency, "req_timeout": req_timeout, "reasoning": reasoning, "structured_output": structured_output, "provider": provider, "probe_first": probe_first, "eval_version": eval_version, - "identity_eval_version": identity_eval_version, "seed_schedule": seed_schedule} + "identity_eval_version": identity_eval_version, "seed_schedule": seed_schedule, + "rating_rubric": rating_rubric, "request_metadata": request_metadata, + "reuse_valid_records": reuse_valid_records} + prior_answers = {} + if reuse_valid_records and rpath.exists(): + for line in rpath.read_text().splitlines(): + record = json.loads(line) + if (record.get("event") == "answer_parsed" and record.get("parsed") is True + and record.get("protocol_id") == protocol_id): + key = (record["item_id"], record["sample"], tuple(record["presented_order"])) + prior_answers[key] = record _append_record(rpath, {"event": "run_started", "run_id": run_id, "protocol_id": protocol_id, "settings": settings, "items": items, "planned_requests": len(plan)}) @@ -287,6 +356,13 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp "presented_options": req["presented_options"], "presented_order": req["perm"], "sample": req["sample"], "prompt": req["prompt"], "settings": settings, "eval_version": eval_version, "seed": seed} + prior_key = (item["id"], req["sample"], tuple(req["perm"])) + if prior_key in prior_answers: + prior = prior_answers[prior_key] + _append_record(rpath, {"event": "request_reused", **request_meta, + "source_run_id": prior["run_id"]}) + return {"text": prior["text"], "rescued": prior.get("rescued", False), + "error": None, "reused": True} payload = {"model": model, "messages": [{"role": "user", "content": req["prompt"]}], "temperature": temperature, "n": req["cnt"], "max_tokens": max_tokens} if seed is not None: @@ -303,7 +379,7 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp try: _append_record(rpath, {"event": "request_started", "phase": phase, **request_meta, "payload": payload}) - data = await asyncio.wait_for(openrouter_request(payload), timeout=req_timeout) + data = await asyncio.wait_for(request_fn(payload), timeout=req_timeout) _append_record(rpath, {"event": "request_completed", "phase": phase, **request_meta, "response": data, "provider": data.get("provider"), "usage": data.get("usage")}) @@ -322,6 +398,8 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp or "(thinking truncated)"}, {"role": "user", "content": _force_msg(item["n"])}, ]} + if seed is not None: + rescue_payload["seed"] = seed if response_format is not None: rescue_payload["response_format"] = response_format if reasoning is not None: @@ -332,7 +410,8 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp **request_meta, "payload": rescue_payload, "initial_response_message": message}) rescue = await _force_answer(model, req["prompt"], message, temperature, - max_tokens, req_timeout, reasoning, response_format, item["n"], provider) + max_tokens, req_timeout, reasoning, response_format, + item["n"], provider, seed, request_fn) _append_record(rpath, {"event": "request_completed", "phase": phase, **request_meta, "response": rescue, "provider": rescue.get("provider"), "usage": rescue.get("usage")}) @@ -359,9 +438,10 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp return [first] + await asyncio.gather(*(call(seq, req) for seq, req in enumerate(plan[1:], start=1))) results = asyncio.run(run_all()) - agg = {i: {"p_samples": [], "texts": [], "failed": 0, "rescued": 0, "prompt": ""} + agg = {i: {"p_samples": [], "rating_samples_raw": [], "rating_samples_transformed": [], + "texts": [], "failed": 0, "rescued": 0, "prompt": ""} for i in range(len(items))} - for req, result in zip(plan, results): + for seq, (req, result) in enumerate(zip(plan, results)): i, n, perm = req["i"], items[req["i"]]["n"], req["perm"] agg[i]["prompt"] = req["prompt"] agg[i]["rescued"] += int(result["rescued"]) @@ -373,21 +453,29 @@ def read_items_rated(model: str, items: list[dict], *, n_samples: int = 12, temp text = result["text"] agg[i]["texts"].append(text) rated = _parse_ratings(text, n) - _append_record(rpath, {"event": "answer_parsed", "run_id": run_id, "protocol_id": protocol_id, + _append_record(rpath, {"event": "answer_parsed", "request_id": f"{run_id}_{seq:03d}", + "run_id": run_id, "protocol_id": protocol_id, "model": model, "item_id": items[i]["id"], "sample": req["sample"], - "presented_order": perm, "text": text, "parsed": rated is not None}) + "presented_order": perm, "text": text, "rescued": result["rescued"], + "parsed": rated is not None}) if rated is None: continue - r_canon = np.zeros(n) + r_raw = np.zeros(n) for j in range(n): - r_canon[perm[j]] = rated[j] - agg[i]["p_samples"].append(r_canon / r_canon.sum()) + r_raw[perm[j]] = rated[j] + r_transformed = 6 - r_raw if rating_rubric == "reversed" else r_raw.copy() + agg[i]["rating_samples_raw"].append(r_raw) + agg[i]["rating_samples_transformed"].append(r_transformed) + agg[i]["p_samples"].append(r_transformed / r_transformed.sum()) out = [] for i, item in enumerate(items): ps = agg[i]["p_samples"] p = np.mean(ps, axis=0) if ps else np.full(item["n"], np.nan) row = {"id": item["id"], "p": p, "p_samples": [x.tolist() for x in ps], + "rating_samples_raw": [x.tolist() for x in agg[i]["rating_samples_raw"]], + "rating_samples_transformed": [x.tolist() for x in agg[i]["rating_samples_transformed"]], + "rating_rubric": rating_rubric, "pmass_allowed": len(ps) / n_samples, "n_samples": n_samples, "valid_samples": len(ps), "failed_samples": agg[i]["failed"], "rescued_samples": agg[i]["rescued"], "prompt": agg[i]["prompt"],