| Benchmark | Model | Rank | Results |
|---|---|---|---|
| arithmetic-reasoning-on-gsm8k | CGS (Qwen2.5-1.5B → Qwen2.5-32B) | – | Accuracy: 92.5 |
| mathematical-reasoning-on-aime24 | CGS (Qwen2.5-7B → Qwen2.5-72B) | – | Acc: 15.6 |
| mathematical-reasoning-on-minerva-math | CGS (Qwen2.5-7B → Qwen2.5-72B) | – | Accuracy: 63.6 |