Werewolf Benchmark - Win Rate: leaderboard
Win rate sub-leaderboard from the Werewolf Benchmark, tracking overall win percentage across both wolf and villager roles.
Metric: Win Rate (%). Source: werewolf.foaster.ai. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 95.6 | #98 |
| 2 | Gemini 2.5 Pro | 58.9 | #152 |
| 3 | Gemini 2.5 Flash | 55.6 | #245 |
| 4 | Grok 4 Fast (Reasoning) | 53.3 | #254 (Grok 4 Fast) |
| 5 | Grok 4 | 51.1 | #181 |
| 6 | Kimi K2 0905 | 46.7 | #288 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 44.4 | #300 |
| 8 | GPT-5 Mini | 38.9 | #186 |
| 9 | Kimi K2 | 38.9 | #243 |
| 10 | GPT-OSS-120B | 16.7 | #341 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=werewolf-benchmark-win-rate · How It Works · Data refreshed daily, snapshot 2026-10-20.