RevengeBench - Poker: leaderboard

RevengeBench asks an agent to reconstruct a hidden game bot's policy from observed behavior and experiments. This series reports Poker distance reduction relative to a starter policy, in percent, under mini-SWE-agent with the best of five rounds reported. It measures how closely the recovered policy matches held-out actions; higher is better and negative values mean worse than the starter.

Metric: Distance reduction (%; best round, mini-SWE-agent). Source: revengebench.com. Saturation forecast: Around June 2027. 12 models tracked. Updated 2 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.582.2#50
2GLM-5.178#195
3GPT-576.3#177
4Qwen 3.6 35B A3B74.4#484
5DeepSeek V4 Flash73.6#265
6DeepSeek V4 Pro73#198
7Gemma 4 31B68.5#385
8Gemma 4 26B A4B63.6#581
9Grok 4.1 Fast60.6#429
10Kimi K2.658.4#191
11GPT-5.4 Mini44.2#401
12GPT-OSS-120B40.5#707

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=revengebench-poker · How It Works · Data refreshed daily, snapshot 2026-11-17.