Video Question Answering on LVBench (test)
48.9AccuracyGPT-4o
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4oSize=-, # Frames=1fps2026.04 | 48.9 | |
| InternVL 2.5Size=78B, # Frames=642026.04 | 43.6 | |
| VideoMindSize=7B, # Frames=322026.04 | 40.8 | |
| SVAgentBackbone=Qwen3-VL, Size=8B, # Frames=82026.04 | 40.8 | |
| SVAgentBackbone=Qwen2.5-VL, Size=7B, # Frames=82026.04 | 40.6 | |
| Qwen2.5-VLSize=72B, # Frames=32 / 768∗2026.04 | 39.3 | |
| LLaVA-VideoSize=72B, # Frames=32 / 64∗2026.04 | 38.7 | |
| SVAgentBackbone=Qwen2.5-VL, Size=3B, # Frames=82026.04 | 38.5 | |
| SVAgentBackbone=Qwen3-VL, Size=4B, # Frames=82026.04 | 37.6 | |
| Qwen2.5-VLSize=7B, # Frames=82026.04 | 33.9 | |
| Qwen3-VLSize=8B, # Frames=82026.04 | 33.4 | |
| Gemini 1.5 ProSize=-, # Frames=1fps2026.04 | 33.1 | |
| Qwen3-VLSize=4B, # Frames=82026.04 | 32.1 | |
| Qwen2.5-VLSize=3B, # Frames=82026.04 | 31.6 |