| code-generation-on-livecodebench-v6 | JudgeRLVR (Qwen3-30B-A3B) | – | Pass@1: 63.9 |
| logic-puzzle-solving-on-zebralogic | Mixed Strategy (Qwen3-30B-A3B) | – | Accuracy: 87.4 |
| mathematical-reasoning-on-aime-2025 | Mixed Strategy (Qwen3-30B-A3B) | – | Accuracy: 79.7 |
| mathematical-reasoning-on-aime24 | JudgeRLVR (Qwen3-30B-A3B) | – | Acc: 89.0 |
| mathematical-reasoning-on-hmmt-feb-2025 | JudgeRLVR (Qwen3-30B-A3B) | – | Accuracy: 70.0 |
| mathematical-reasoning-on-math-500 | Vanilla RLVR (Qwen3-30B-A3B) | – | Accuracy: 98.0 |
| multiple-choice-question-answering-mcqa-on-gpqa-diamond | Judge Only (Qwen3-30B-A3B) | – | Accuracy: 66.9 |