Long Video Understanding on VideoMME Overall (w/o sub.) 2025
75AccuracyGemini-1.5-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-1.5-ProSize=-, Frames=0.5 fps2025.08 | 75 | |
| GPT-4oSize=-, Frames=0.5 fps / 3842025.08 | 71.9 | |
| Qwen2.5-VL-7BSize=7B, Frames=7682025.08 | 65.1 | |
| Video-MTRSize=7B, Frames=802025.08 | 62.7 | |
| Video-MTRSize=7B, Frames=642025.08 | 62.2 | |
| Video-R1Size=7B, Frames=642025.08 | 61.4 | |
| Qwen2.5-VL-7BSize=7B, Frames=802025.08 | 59.5 | |
| Video-R1Size=7B, Frames=322025.08 | 59.3 | |
| Video-MTRSize=7B, Frames=322025.08 | 59 | |
| Qwen2.5-VL-7BSize=7B, Frames=642025.08 | 58.4 | |
| LLaVA-OneVisionSize=7B, Frames=322025.08 | 58.2 | |
| VideoMemAgentSize=-, Frames=722025.08 | 57.4 | |
| VideoAgentSize=-, Frames=872025.08 | 56 | |
| Video-XLSize=7B, Frames=128 / 2562025.08 | 55.5 | |
| Qwen2.5-VL-7BSize=7B, Frames=322025.08 | 53.6 | |
| LongVASize=7B, Frames=128 / 2562025.08 | 52.6 | |
| DrVideo(GPT-4)Size=-, Frames=0.2/0.5 fps2025.08 | 51.7 | |
| Video-CCAMSize=9B, Frames=962025.08 | 50.3 | |
| VideoLLaMA2Size=8×7B, Frames=82025.08 | 47.9 | |
| Video-LLaVASize=7B, Frames=82025.08 | 39.9 | |
| VideoChat2Size=7B, Frames=162025.08 | 39.5 |