Video Understanding on WorldSense (test)
42.6Overall AccuracyGPT-4o
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-4oCategory=Proprietary Models2026.05 | 42.6 | — | — | — | |
| VISDCategory=Ours2026.05 | 39.9 | 39.4 | 40.4 | 40.5 | |
| VisionCoach-7BCategory=Tool-free Methods, Evaluated by Authors=true2026.05 | 39.2 | 38.4 | 40.2 | 39.6 | |
| Open-o3-Video-7BCategory=Tool-free Methods, Evaluated by Authors=true2026.05 | 38.9 | 37.4 | 39.7 | 40.5 | |
| VideoRFT-7BCategory=Tool-free Methods, WorldSense Evaluated by Authors=true2026.05 | 38.4 | 36.5 | 40 | 40.1 | |
| VideoR1-7BCategory=Tool-free Methods, WorldSense Evaluated by Authors=true2026.05 | 36.9 | 35.3 | 37.7 | 39 | |
| LongVT-RL-7BCategory=Tool-calling Methods, Evaluated by Authors=true2026.05 | 36.1 | 33.1 | 39.1 | 37.8 | |
| Qwen2.5-VL-7BCategory=Tool-free Methods, WorldSense Evaluated by Authors=true2026.05 | 35.5 | 32.9 | 38.4 | 37 |