Poker Agent
Games · 2025-12-23
Poker Agent evaluates probabilistic reasoning and strategy in 10-player no-limit Texas Hold'em tournaments. This row records the normalized TrueSkill rating derived from chip-count rankings and elimination order; higher ratings indicate better performance.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.2 | 1132 |
| GPT-5 | 1103 |
| Gemini 3 Flash Preview | 1100 |
| DeepSeek-V3.2 | 1090 |
| Grok 4.1 Fast | 1079 |
| Gemini 3 Pro Preview | 1079 |
| DeepSeek-V3.1 | 1058 |
| Sonnet 4.5 | 1056 |
| GPT-5.1 | 1039 |
| Grok 4 Fast | 1034 |
| Opus 4.5 | 1033 |
| Gemini 2.5 Pro | 1033 |
| gpt-oss-120b | 1015 |
| Kimi K2 Thinking | 1012 |
| Qwen3 Max Preview | 995 |
| GLM 4.6 | 946 |
| Llama 4 Maverick Instruct | 891 |