| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-olympiadbench | Qwen2.5-32B + RL (synthetic) | #38 | Accuracy: 44.53 |
| mathematical-reasoning-on-math-500 | Qwen2.5-32B + RL (synthetic) | #83 | Accuracy: 82.8 |
| mathematical-reasoning-on-aime-2025 | Qwen2.5-32B + RL (synthetic) | #140 | Accuracy: 12.5 |