Video Question Answering on MLVU (test)
75.7AccuracyInternVL 2.5
Evaluation results
| Method | Links | |
|---|---|---|
| InternVL 2.5Size=78B, # Frames=642026.04 | 75.7 | |
| Video-RAGSize=72B, # Frames=642026.04 | 73.8 | |
| Qwen2.5-VLSize=72B, # Frames=32 / 768∗2026.04 | 73.1 | |
| LLaVA-VideoSize=72B, # Frames=32 / 64∗2026.04 | 72.9 | |
| VgentSize=7B, # Frames=3202026.04 | 72.1 | |
| SVAgentBackbone=Qwen3-VL, Size=8B, # Frames=82026.04 | 65.6 | |
| SVAgentBackbone=Qwen3-VL, Size=4B, # Frames=82026.04 | 64.9 | |
| GPT-4oSize=-, # Frames=1fps2026.04 | 64.6 | |
| VideoMindSize=7B, # Frames=322026.04 | 64.4 | |
| Gemini 1.5 ProSize=-, # Frames=1fps2026.04 | 64 | |
| SVAgentBackbone=Qwen2.5-VL, Size=7B, # Frames=82026.04 | 62.7 | |
| SVAgentBackbone=Qwen2.5-VL, Size=3B, # Frames=82026.04 | 61.2 | |
| Qwen2.5-VLSize=7B, # Frames=82026.04 | 54.7 | |
| Qwen3-VLSize=8B, # Frames=82026.04 | 54.5 | |
| Qwen2.5-VLSize=3B, # Frames=82026.04 | 53.6 | |
| Qwen3-VLSize=4B, # Frames=82026.04 | 53.4 |