| Benchmark | Model | Rank | Results |
|---|---|---|---|
| multimodal-reasoning-on-mathvista | Evidence-RL (Qwen3.5-9B) | – | Accuracy: 83.75 |
| multimodal-reasoning-on-mmmu | Evidence-RL (Qwen3.5-9B) | – | Accuracy: 75.67 |
| object-counting-on-countbench | Evidence-RL (Qwen3-VL-8B) | – | Accuracy: 95.92 |