RevengeBench - RoboCode: leaderboard

RevengeBench asks an agent to reconstruct a hidden game bot's policy from observed behavior and experiments. This series reports RoboCode 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-573.3#177
2GPT-5.570.7#50
3GLM-5.156.8#195
4Grok 4.1 Fast48#429
5Kimi K2.644.6#191
6DeepSeek V4 Pro44.1#198
7GPT-5.4 Mini43#401
8Gemma 4 26B A4B37.9#581
9Gemma 4 31B33.7#385
10DeepSeek V4 Flash28.6#265
11Qwen 3.6 35B A3B23.2#484
12GPT-OSS-120B18.1#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-robocode · How It Works · Data refreshed daily, snapshot 2026-11-17.