#ModelAccuracyPaper
1TASKER (ours) Qwen3-VL79.40link
2VideoAgent [52] (Qwen3-VL)75.80link
3TASKER (Ours) GPT-4o68.60link
4LVNet68.20link
5TASKER (Ours) GPT-468.00link
6LangRepo66.20link
7VideoTree66.20link
8LifelongMemory64.10link
9LLoVi61.20link
10MVU60.30link
11VideoAgent [52]60.20link
12IG-VLM59.80link
13ProViQ57.10link
14MoReVQA51.70link
15Vamos51.20link
16Video-LLA+INTP38.60link
17VisualClaw: A Real-Time, Personalized Agent for the Physical World3.85link
EgoSchema egoschema Leaderboard