Long Video Understanding on LongVideoBench 23sec-60 min
74.4AccuracySeed1.5-VL
Evaluation Results
| Method | Links | |
|---|---|---|
| Seed1.5-VLSize=8B, Input Frames=>=642025.12 | 74.4 | |
| GPT-4oInput Frames=>=642025.12 | 66.7 | |
| Eagle2.5Size=8B, Input Frames=>=642025.12 | 66.4 | |
| VideoMem*Size=8B, Input Frames=642025.12 | 63.3 | |
| InternVL3.5Size=8B, Input Frames=>=642025.12 | 62.1 | |
| Video-XL2Size=8B, Input Frames=>=642025.12 | 61 | |
| Flow4AgentSize=7B, Input Frames=>=642025.12 | 60.4 | |
| VideoNSASize=7B, Input Frames=>=642025.12 | 60 | |
| VideoLLaMA3Size=7B, Input Frames=>=642025.12 | 59.8 | |
| GPT-4VInput Frames=>=642025.12 | 59.1 | |
| Qwen3-VL*Size=8B, Input Frames=642025.12 | 59.1 | |
| ApolloSize=7B, Input Frames=>=642025.12 | 58.5 | |
| LLaVA-videoSize=7B, Input Frames=>=642025.12 | 58.2 | |
| NVILASize=8B, Input Frames=>=642025.12 | 57.7 | |
| LLaVA-OnevisionSize=7B, Input Frames=>=642025.12 | 56.4 | |
| Qwen2.5-VLSize=7B, Input Frames=>=642025.12 | 54.7 | |
| LLaVA-Next-VideoSize=7B, Input Frames=>=642025.12 | 50.5 | |
| ShareGPT4VideoSize=8B, Input Frames=>=642025.12 | 39.7 | |
| Video-LLaVASize=7B, Input Frames=>=642025.12 | 39.1 |