| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-minerva-math | ReSPO (Qwen3-30B-A3B-Base, N=16) | #16 | Accuracy: 45.7 |
| mathematical-reasoning-on-olympiadbench | ReSPO (Qwen3-30B-A3B-Base, N=16) | #25 | Accuracy: 51.2 |
| mathematical-reasoning-on-math-500 | ReSPO (Qwen3-30B-A3B-Base, N=16) | #66 | Accuracy: 85.7 |
| mathematical-reasoning-on-aime24 | ReSPO (Qwen3-30B-A3B-Base, N=16) | #99 | Acc: 32.1 |
| mathematical-reasoning-on-aime-2025 | ReSPO (Qwen3-30B-A3B-Base, N=8) | #115 | Accuracy: 25.2 |