| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-mathverse | NPO (early + late-stage) | – | Accuracy: 60.0 |
| multimodal-reasoning-on-mathvision | AutoNPO | – | Accuracy: 55.72 |
| multimodal-reasoning-on-mathvista | AutoNPO | – | Accuracy: 79.2 |
| multimodal-reasoning-on-mmmu-pro | AutoNPO | – | Accuracy: 57.24 |
| vision-language-understanding-on-mmstar | AutoNPO | – | Accuracy: 72.63 |