| Benchmark | Model | Rank | Results |
|---|---|---|---|
| code-generation-on-livecodebench-v6 | SAF (Qwen3-8B) | – | Pass@1: 37.43 |
| mathematical-reasoning-on-aime-2025 | SAF (Qwen3-4B) | – | Accuracy: 53.96 |
| mathematical-reasoning-on-aime24 | SAF (Qwen3-8B) | – | Acc: 64.79 |
| mathematical-reasoning-on-hmmt-feb-2025 | SAF (Qwen3-4B) | – | Accuracy: 31.15 |