Long-form Video Understanding on VideoMME long subset w/ sub
81.2AccuracyVideoSeek
Evaluation results
| Method | Links | |
|---|---|---|
| VideoSeek#Frames=15.92026.03 | 81.2 | |
| GPT-5 (Base)#Frames=3842026.03 | 78.1 | |
| Gemini 1.5 Pro#Frames=1,2332026.03 | 77.4 | |
| GPT-4o#Frames=3842026.03 | 72.1 | |
| DrVideo#Frames=493.22026.03 | 71.7 | |
| Qwen2.5-VL-72B#Frames=2562026.03 | 64.4 |