| code-generation-on-humaneval | Qwen-2.5-32B-R-TAP | – | Pass@1: 97.9 |
| code-generation-on-mbpp | Qwen-2.5-32B-R-TAP | – | Accuracy: 87.9 |
| mathematical-reasoning-on-aime-2025 | Qwen3-235B-A22B-R-TAP (MoE, 235B) | – | Accuracy: 88.7 |
| mathematical-reasoning-on-aime24 | Qwen3-235B-A22B-R-TAP (MoE, 235B) | – | Acc: 90.9 |
| mathematical-reasoning-on-math-500 | MiMo-RL-7B-R-TAP | – | Accuracy: 97.3 |
| mathematical-reasoning-on-minerva-math | R1-Distill-Qwen-7B-R-TAP | – | Accuracy: 42.3 |
| mathematical-reasoning-on-olympiadbench | R1-Distill-Qwen-7B-R-TAP | – | Accuracy: 53.8 |
| multi-task-language-understanding-on-mmlu-pro | MiMo-RL-7B-R-TAP | – | Accuracy: 65.9 |
| multiple-choice-question-answering-mcqa-on-gpqa-diamond | MiMo-RL-7B-R-TAP | – | Accuracy: 60.7 |