| Benchmark | Model | Rank | Results |
|---|---|---|---|
| arithmetic-reasoning-on-gsm8k | OpenMath-Nemotron-14B + iGRPO | – | Accuracy: 94.69 |
| mathematical-reasoning-on-aime-2025 | OpenMath-Nemotron-14B + iGRPO | – | Accuracy: 65.57 |
| mathematical-reasoning-on-aime24 | OpenMath-Nemotron-14B + iGRPO | – | Acc: 76.72 |
| mathematical-reasoning-on-math-500 | OpenMath-Nemotron-14B + iGRPO | – | Accuracy: 96.7 |
| mathematical-reasoning-on-minerva-math | DeepSeek-R1-Distill-Qwen-14B + iGRPO | – | Accuracy: 47.06 |