Video Question Answering on M3-Bench robot
43.7ME ScoreHOMER
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| HOMERMethod Category=Agent Method, Backbone=Gemini32026.07 | 43.7 | 42.4 | 43.3 | 60.4 | 28.1 | 42.8 | |
| M3-Agent-Gemini3Method Category=Agent Method, Backbone=Gemini32026.07 | 38.5 | 37.7 | 35.1 | 48.5 | 27.2 | 37.3 | |
| M3-AgentMethod Category=Agent Method2026.07 | 32.8 | 29.4 | 31.2 | 43.3 | 19.1 | 30.7 | |
| MA-LMMMethod Category=Online Video Understanding2026.07 | 25.6 | 23.4 | 22.7 | 39.1 | 14.4 | 24.4 | |
| M3-Agent-Gemini25Method Category=Agent Method, Backbone=Gemini252026.07 | 24.6 | 40 | 23.3 | 32.1 | 19.9 | 24.6 | |
| Flash-VStreamMethod Category=Online Video Understanding2026.07 | 21.6 | 19.4 | 19.3 | 24.3 | 14.1 | 19.4 | |
| Gemini-GPT-HybridMethod Category=Agent Method2026.07 | 21.3 | 25.5 | 22.7 | 28.8 | 23.1 | 24 | |
| Gemini-AgentMethod Category=Agent Method2026.07 | 15.8 | 17.1 | 15.3 | 20 | 15.5 | 16.9 | |
| MovieChatMethod Category=Online Video Understanding2026.07 | 13.3 | 9.8 | 12.2 | 15.7 | 7 | 11.2 |