Poker Agent: leaderboard
Which model can make the most money playing poker?.
Metric: TrueSkill Rating. Source: vals.ai. Saturation forecast: Around March 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 1131.83 | #110 |
| 2 | GPT-5 | 1103.18 | #98 |
| 3 | Gemini 3 Flash (Preview) | 1100.21 | #82 |
| 4 | DeepSeek V3.2 (Thinking) | 1090.3 | #205 (DeepSeek V3.2) |
| 5 | Grok 4.1 Fast (Reasoning) | 1079.22 | #217 (Grok 4.1 Fast) |
| 6 | Gemini 3 Pro (Preview) | 1078.9 | #66 |
| 7 | DeepSeek V3.1 | 1058.23 | #264 |
| 8 | Claude Sonnet 4.5 | 1055.5 | #142 |
| 9 | GPT-5.1 (2025-11-13) | 1038.59 | #101 |
| 10 | Grok 4 Fast (Reasoning) | 1034.3 | #254 (Grok 4 Fast) |
| 11 | Claude Opus 4.5 (20251101) | 1033.38 | #74 |
| 12 | Gemini 2.5 Pro | 1032.6 | #152 |
| 13 | GPT-OSS-120B | 1015.33 | #341 |
| 14 | Kimi K2 (Thinking) | 1011.63 | #243 (Kimi K2) |
| 15 | Qwen 3 Max (Preview) | 994.51 | #230 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=poker-agent · How It Works · Data refreshed daily, snapshot 2026-10-20.