diff --git a/python/ray/rllib/tuned_examples/pong-ppo.yaml b/python/ray/rllib/tuned_examples/pong-ppo.yaml
new file mode 100644
index 000000000..9fdf5debd
--- /dev/null
+++ b/python/ray/rllib/tuned_examples/pong-ppo.yaml
@@ -0,0 +1,19 @@
+# On a Tesla K80 GPU, this achieves the maximum reward in about 1-1.5 hours.
+#
+# $ python train.py -f tuned_examples/pong-ppo.yaml --num-gpus=1
+#
+# - PPO_PongDeterministic-v4_0:  TERMINATED [pid=16387], 4984 s, 1117981 ts, 21 rew
+# - PPO_PongDeterministic-v4_0:  TERMINATED [pid=83606], 4592 s, 1068671 ts, 21 rew
+#
+pong-deterministic-dqn:
+    env: PongDeterministic-v4
+    run: PPO
+    resources:
+        cpu: 1
+        gpu: 1
+    stop:
+        episode_reward_mean: 21
+    config:
+        gamma: 0.99
+        num_sgd_iter: 20
+        devices: ["/gpu:0"]