Mathematical Reasoning on AIME 2025 (Acceptance length, Delta %)
Evaluation results
| Method | Links | | |
|---|
| TTSTarget Model=Qwen/Qwen3.5-122B2026.05 | | 4.9 | 71.8 |
|---|
| TTSTarget Model=Qwen/Qwen3-4B2026.05 | | 4.7 | 47 |
|---|
| TTSTarget Model=Qwen/Qwen3-8B2026.05 | | 4.2 | 33.1 |
|---|
| DFlashTarget Model=Qwen/Qwen3-4B2026.05 | | 3.2 | — |
|---|
| DFlashTarget Model=Qwen/Qwen3-8B2026.05 | | 3.1 | — |
|---|
| DFlashTarget Model=Qwen/Qwen3.5-122B2026.05 | | 2.9 | — |
|---|