| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-mathverse | LLaDA-V + Bi-VRL (ours) | – | Accuracy: 31.2 |
| multimodal-reasoning-on-mmmu | LLaDA-V + Bi-VRL (ours) | – | Accuracy: 53.4 |
| multimodal-reasoning-on-mmmu-pro | LLaDA-V + Bi-VRL (ours) | – | Accuracy: 38.3 |
| vision-language-understanding-on-mmstar | LLaDA-V + Bi-VRL (ours) | – | Accuracy: 63.2 |
| visual-question-answering-vqa-on-seed-seedbench | LLaDA-V + Bi-VRL (ours) | – | Accuracy: 81.1 |