Long-form Video Understanding on LVBench w/o sub (test)
74.2AccuracyDVD
Evaluation results
| Method | Links | |
|---|---|---|
| DVD#Frames=8,0742026.03 | 74.2 | |
| VideoSeek#Frames=92.32026.03 | 68.4 | |
| MR. Video#Frames=8,0742026.03 | 60.8 | |
| GPT-5 (Base)#Frames=3842026.03 | 60.1 | |
| Gemini 2.0 Flash#Frames=4,0372026.03 | 48.3 | |
| Qwen2.5-VL-72B#Frames=7682026.03 | 47.3 | |
| VCA#Frames=20.02026.03 | 41.3 | |
| Gemini 1.5 Pro#Frames=3,6002026.03 | 33.1 | |
| GPT-4o#Frames=3842026.03 | 30.8 | |
| VideoAgent#Frames=25.52026.03 | 29.3 | |
| VideoTree#Frames=103.22026.03 | 28.8 |