Mathematical Reasoning on AIME 2025 (Pass@5 and Token Count)
86.67Pass@50-CoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| 0-CoTModel=Qwen 3 32B2025.11 | 86.67 | 15,137 | |
| PSModel=Qwen 3 32B2025.11 | 86.67 | 14,644 | |
| 0-CoTModel=Qwen 3 4B2025.11 | 83.33 | 16,293 | |
| PSModel=Qwen 3 14B2025.11 | 83.33 | 15,701 | |
| CoReModel=Qwen 3 32B2025.11 | 83.33 | 15,456 | |
| PSModel=Qwen 3 4B2025.11 | 80 | 16,115 | |
| 0-CoTModel=Qwen 3 14B2025.11 | 80 | 16,185 | |
| CoReModel=Qwen 3 14B2025.11 | 80 | 16,086 | |
| CoReModel=Qwen 3 4B2025.11 | 76.67 | 16,695 | |
| F-CoTModel=Qwen 3 14B2025.11 | 76.67 | 12,788 | |
| F-CoTModel=Qwen 3 32B2025.11 | 76.67 | 11,143 | |
| F-CoTModel=Qwen 3 4B2025.11 | 73.33 | 11,287 | |
| PSModel=Qwen 3 0.6B2025.11 | 26.67 | 14,289 | |
| 0-CoTModel=Qwen 3 0.6B2025.11 | 23.33 | 14,904 | |
| CoReModel=Qwen 3 0.6B2025.11 | 23.33 | 14,035 | |
| F-CoTModel=Qwen 3 0.6B2025.11 | 23.33 | 8,460 |