| multimodal-reasoning-on-babyvision | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 32.22 |
| multimodal-reasoning-on-mathvision | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 83.71 |
| multimodal-reasoning-on-mathvista | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 89 |
| multimodal-reasoning-on-mmmu | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 82.89 |
| multimodal-reasoning-on-mmmu-pro | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 76.47 |
| video-question-answering-on-lvbench | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 59.07 |
| video-question-answering-on-mlvu | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 82.52 |
| video-question-answering-on-mmvu | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 68.9 |
| video-question-answering-on-mvbench | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Avg.: 73.31 |
| video-question-answering-on-video-mmmu | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 80 |
| video-question-answering-on-videommmu | Qwen3.5-397B-A17B | – | Accuracy: 80.0 |
| vision-language-understanding-on-mmstar | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 80.67 |
| visual-question-answering-vqa-on-mmbench-cn | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 91.72 |
| visual-question-answering-vqa-on-mmbench-en | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 91.56 |
| visual-question-answering-vqa-on-realworldqa | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | – | Accuracy: 81.7 |