Long-video understanding on M3-Bench web
55.4MDRPyraVid
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PyraVidModel Size=32B2026.05 | 55.4 | 42.9 | 54.5 | 70.8 | 51 | 56.3 | |
| Gemini-2.0-FlashVariant=Flash2026.05 | 52.5 | 30.6 | 34.1 | 46 | 51 | 44.9 | |
| PyraVidModel Size=8B2026.05 | 52.5 | 32.7 | 43.2 | 67.9 | 45.1 | 51.1 | |
| Qwen3-VL-8B-InstructModel Size=8B, Variant=Instruct2026.05 | 46.5 | 26.5 | 27.3 | 46.7 | 29.4 | 35.7 | |
| M3-AgentModel Size=32B, Variant=RL2026.05 | 45.9 | 28.4 | 44.3 | 59.3 | 53.9 | 48.9 | |
| MA-LMM2026.05 | 26.8 | 10.5 | 22.4 | 39.3 | 15.8 | 24.3 | |
| Flash-VStream2026.05 | 24.5 | 10.3 | 24.6 | 32.5 | 20.2 | 23.6 | |
| MovieChat2026.05 | 12.2 | 6.6 | 12.5 | 17.4 | 11.1 | 12.6 |