Video Quality Understanding on Q-Bench-Video (dev)
76.78Yes-or-No AccGemini-3-Pro
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Gemini-3-ProRequires VQA-labeled data=false, Proprietary=true2026.01 | 76.78 | 61 | 47.87 | 61 | 69.56 | 59.49 | 60 | 63.02 | |
| Q-Router (GPT-4o)Requires VQA-labeled data=true, Reasoning backbone=GPT-4o2026.01 | 76 | 57.01 | 43.33 | 59.31 | 65.31 | 61.46 | 50.22 | 60.07 | |
| OmniVQA-ChatterRequires VQA-labeled data=true2026.01 | 75.51 | 59.76 | 40.37 | 62.05 | 61.58 | 63.45 | 42.38 | 59.08 | |
| VQA^2Requires VQA-labeled data=true2026.01 | 73.81 | 56.4 | 38.33 | 60.7 | 56.65 | 61.09 | 38.11 | 56.67 | |
| QualiRAGRequires VQA-labeled data=false2026.01 | 73.18 | 62.95 | 44.95 | 61.34 | 70.74 | 61.69 | 48.17 | 61.66 | |
| VQAThinkerRequires VQA-labeled data=true2026.01 | 70.42 | 56.56 | 35.15 | 55.95 | 64.56 | 61.69 | 30.91 | 55.7 | |
| GPT-5Requires VQA-labeled data=false, Proprietary=true2026.01 | 68.72 | 56.49 | 43.21 | 56.58 | 68.99 | 59.91 | 37.42 | 57.22 | |
| Qwen2.5-VL-7BRequires VQA-labeled data=false2026.01 | 68.54 | 57.01 | 35.56 | 54.94 | 63.12 | 57.27 | 35.7 | 55.3 | |
| GLM-4.1V-9BRequires VQA-labeled data=false2026.01 | 63.85 | 47.96 | 31.45 | 49.28 | 60.77 | 50.59 | 24.25 | 49.15 | |
| Qwen3-VL-8BRequires VQA-labeled data=false2026.01 | 62.44 | 51.81 | 38.92 | 51.68 | 62.85 | 54.9 | 31.94 | 52.11 | |
| LLaVA-OneVision-7BRequires VQA-labeled data=false2026.01 | 62.13 | 52.23 | 38.56 | 48.74 | 61.53 | 48.81 | 44.57 | 52.12 | |
| InternVL3.5-8BRequires VQA-labeled data=false2026.01 | 61.97 | 51.13 | 35.31 | 49.28 | 65.17 | 54.33 | 29.14 | 50.7 | |
| mPLUG-Owl3-7BRequires VQA-labeled data=false2026.01 | 60.82 | 56.52 | 35.84 | 51.34 | 60.46 | 54.26 | 37.3 | 52.44 | |
| InternVL3-9BRequires VQA-labeled data=false2026.01 | 56.57 | 51.58 | 30.4 | 46.35 | 59.8 | 50.83 | 31.51 | 47.61 |