Fail fast on incomplete WVS API panels

Co-Authored-By: PI[gpt-5.6-terra] <288921227+claudypoo@users.noreply.github.com>

scripts/wvs_api/01_gpt5_nano_diagnostic.sh

@@ -2,7 +2,8 @@
 set -eu
 uv run --with 'datasets>=4.0,<5' python scripts/wvs_map.py \
   --api-models openai/gpt-5-nano \
-  --api-disable-reasoning \
+  --api-reasoning-effort low \
   --api-structured-output \
+  --api-require-complete \
   --api-concurrency 1 \
   --out slop/research/wvs/20260917_gpt-5-nano_diagnostic.png

scripts/wvs_map.py

@@ -250,6 +250,8 @@ def main() -> None:
                                  help="send a mandatory model's catalog-supported minimum reasoning effort")
     ap.add_argument("--api-structured-output", action="store_true",
                     help="request a strict rating JSON schema only for a catalog-confirmed supporting model")
+    ap.add_argument("--api-require-complete", action="store_true",
+                    help="exit nonzero rather than render after an explicitly requested API panel is incomplete")
     ap.add_argument("--max-think-tokens", type=int, default=64)
     ap.add_argument("--device", default="cuda" if torch.cuda.is_available() else "cpu")
     ap.add_argument("--out", default="docs/img/wvs/wvs_map_iw.png")
@@ -354,7 +356,10 @@ def main() -> None:
                                 records_path=args.records, verbose_first=True)
         incomplete = [row["id"] for row in rows if row["valid_samples"] != args.api_samples]
         if incomplete:
-            logger.warning(f"{key}: incomplete items {incomplete}; raw evidence is in {args.records}; not cached or plotted")
+            message = f"{key}: incomplete items {incomplete}; raw evidence is in {args.records}; not cached or plotted"
+            logger.warning(message)
+            if args.api_require_complete:
+                raise RuntimeError(message)
             continue
         psamples = {row["id"]: np.array(row["p_samples"]) for row in rows}
         models[key] = model_coord_ci(psamples, resolved, rng)

scripts/wvs_priority_manifest.py

@@ -133,6 +133,8 @@ def main() -> None:
         "|---|---:|---:|---:|---|---:|---|",
         row(catalog["openai/gpt-5-nano"], "required cheapest new 144-call diagnostic"),
         "",
+        "This model's provider rejects `reasoning.enabled=false`; its retry uses `reasoning.effort=low` with strict structured output and fails nonzero unless all 144 samples are valid.",
+        "",
         "## Priority manifest after diagnostic pass",
         "",
         "The order is Grok, OpenAI, Google, then the requested Muse points. `Flash` entries are retained because the user excluded `Fast`, not `Flash`.",
@@ -152,7 +154,7 @@ def main() -> None:
     for name, models in groups:
         lines.append(f"| **{name}** | | | | | | |")
         for model in models:
-            rationale = "clean new full attempt" if model["id"] == "x-ai/grok-4.5" else "new direct panel"
+            rationale = "fresh 144-sample run, separate from incomplete attempts" if model["id"] == "x-ai/grok-4.5" else "new direct panel"
             lines.append(row(model, rationale))
 
     lines.extend([