Video Question Answering on Video-MME v1.0 (test)
65.3Accuracy (Long)GPT-4o
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-4o2026.05 | 65.3 | — | 70.3 | 71.9 | |
| LongVUFrames/FPS=400/1fps2026.05 | 60.6 | — | 59.5 | — | |
| MUSEG-7BModel size=7B2025.05 | 53.6 | 71.9 | 63.7 | 63.1 | |
| Qwen2.5-VL-7B-InstructModel size=7B, Variant=Instruct2025.05 | 52.6 | 71.8 | 62.7 | 62.4 | |
| LLaVA-OV-7B + MuKVFrames/FPS=0.5fps, # Mem. Tok=5.9K2026.05 | 52.1 | — | 57.9 | 61.2 | |
| Qwen3-VL-4B + MuKVFrames/FPS=0.5fps, # Mem. Tok=5.9K2026.05 | 51 | — | 61.8 | 63.6 | |
| LLaVA-OV-7B + StreamMemFrames/FPS=0.5/0.2 fps, # Mem. Tok=6K2026.05 | 50.1 | — | 56.6 | 59.4 | |
| Qwen2.5-VL-3B + MuKVFrames/FPS=0.5 fps, # Mem. Tok=5.9K2026.05 | 50 | — | 61 | 60.8 | |
| Qwen3-VL-2B + MuKVFrames/FPS=0.5fps, # Mem. Tok=5.9K2026.05 | 49.8 | — | 60.8 | 62.6 | |
| DispiderFPS=1 fps2026.05 | 49.7 | — | 53.7 | 57.2 | |
| Qwen3-VL-4BFrames/FPS=7682026.05 | 49.5 | — | 60.2 | 62.5 | |
| Qwen2.5-VL-3B + StreamMemFrames/FPS=4.0/0.5 fps, # Mem. Tok=6K2026.05 | 49.1 | — | 60.1 | 59.5 | |
| LLaVA-OV-7B + LiveVLMFrames/FPS=0.5/0.2 fps2026.05 | 48.8 | — | 56.4 | 57.3 | |
| LLaVA-OV-0.5B + MuKVFrames/FPS=0.5fps, # Mem. Tok=5.9K2026.05 | 48.5 | — | 41.4 | 49.1 | |
| Qwen3-VL-2BFrames/FPS=7682026.05 | 48.5 | — | 59 | 61.5 | |
| LLaVA-OV-0.5B + ReKVFrames/FPS=0.5 fps, # Mem. Tok=5.9K2026.05 | 47.9 | — | 40.1 | 48.2 | |
| Qwen2.5-VL-3BFrames/FPS=7682026.05 | 47.2 | — | 58 | 60.3 | |
| LongVAFrames=1282026.05 | 46.2 | — | 50.4 | 52.6 | |
| LLaVA-OV-7BFrames/FPS=322026.05 | 46.2 | — | 54.7 | 56.9 | |
| LLaVA-OV-0.5BFrames/FPS=322026.05 | 46.2 | — | 39.7 | 45.4 | |
| MovieChat+Frames=20482026.05 | 33.4 | — | — | 38.2 | |
| Qwen2.5-VL-3B + InfiniPot-VFrames/FPS=768, # Mem. Tok=6K2026.05 | — | — | — | 59.3 |