Long Video Understanding on LVBench Overall 2024
48.9AccuracyGPT-4o
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4oSize=-, Frames=0.5 fps / 3842025.08 | 48.9 | |
| Qwen2.5-VL-7BSize=7B, Frames=7682025.08 | 45.3 | |
| Video-MTRSize=7B, Frames=802025.08 | 42.3 | |
| Video-MTRSize=7B, Frames=642025.08 | 41.8 | |
| Video-MTRSize=7B, Frames=322025.08 | 38.2 | |
| Video-R1Size=7B, Frames=642025.08 | 38 | |
| LongVASize=7B, Frames=128 / 2562025.08 | 37.9 | |
| Video-R1Size=7B, Frames=322025.08 | 35.9 | |
| Qwen2.5-VL-7BSize=7B, Frames=642025.08 | 33.7 | |
| Qwen2.5-VL-7BSize=7B, Frames=802025.08 | 33.6 | |
| Gemini-1.5-ProSize=-, Frames=0.5 fps2025.08 | 33.1 | |
| Qwen2.5-VL-7BSize=7B, Frames=322025.08 | 30.3 |