Long Video Understanding on EgoSchema (test)
72.2AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o2024.10 | 72.2 | |
| Gemini-1.5 Pro2024.10 | 72.2 | |
| GPT-4o2026.03 | 72.2 | |
| Gemini 1.5-Pro2026.03 | 71.2 | |
| NVILA-8B-Video + AutoGazeMax #F=1024, Max Res.=35842026.03 | 66.9 | |
| ARIA2024.10 | 66.8 | |
| Gemini-1.5 Flash2024.10 | 65.7 | |
| Qwen2.5-VL-7BMax #F=48, Max Res.=8962026.03 | 65 | |
| VideoLLaMA3-7BMax #F=180, Max Res.=3842026.03 | 63.3 | |
| LLaVA-OneVision-7B + ST-GridPoolInput frames=32, Pooling Strategy=ST-GridPool2026.05 | 62.1 | |
| LLaVA-OV-8BMax #F=32, Max Res.=3842026.03 | 60.1 | |
| LLaVA-OneVision-7BInput frames=322026.05 | 60.1 | |
| InternVideo2-7BBackbone=7B2026.05 | 60 | |
| LLaVA-Video-7B + ST-GridPoolInput frames=64, Pooling Strategy=ST-GridPool2026.05 | 57.8 | |
| LLaVA-Video-7BInput frames=642026.05 | 57.3 | |
| VideoLLaMA2.1-7BBackbone=7B2026.05 | 53.1 |