Multi-Agent Strategic Reasoning on LeducHoldem OOD
83.51First-mover Normalized ScoreGemini-2.5-flash
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Gemini-2.5-flashModel Source=Closed-source, Opponent=NE Bot2026.05 | 83.51 | 96.96 | |
| GPT-5-miniModel Source=Closed-source, Opponent=NE Bot2026.05 | 80.6 | 95.08 | |
| Strat-Reasoner-4BModel Source=Open-source, Opponent=NE Bot2026.05 | 70.12 | 66.64 | |
| Qwen3-32BModel Source=Open-source, Opponent=NE Bot2026.05 | 67.6 | 67.51 | |
| MARSHAL-4BModel Source=Open-source, Opponent=NE Bot2026.05 | 66.42 | 69.22 | |
| Qwen3-4BModel Source=Open-source, Opponent=NE Bot2026.05 | 61.35 | 56.18 | |
| Qwen3-8BModel Source=Open-source, Opponent=NE Bot2026.05 | 60.93 | 64.41 | |
| Gemma3-12BModel Source=Open-source, Opponent=NE Bot2026.05 | 60.55 | 60.13 | |
| SPIRAL-4BModel Source=Open-source, Opponent=NE Bot2026.05 | 56.21 | 55.85 |