Share your thoughts, 1 month free Claude Pro on usSee more

Unseen UI Generalization on OSExpert-Eval (Execution Time)

16Execution Time (Tableau, s)

Human Expert

Updated 4mo ago

Evaluation Results

Method	Links
Human Expert 2026.03		16	48
OSExpert (Explore w/ Qwen-3-VL-8B) 2026.03		28	41
OSExpert (Exploration w/ GPT-5 & UI-TARS) 2026.03		28	44
OSExpert (Explore w/ Qwen-3-VL-8B, wo/ Fast Planner) 2026.03		34	49
OSExpert (Explore w/ Qwen-3-VL-8B, wo/ Boundary Check) 2026.03		75	84
Qwen-3-VL-8B 2026.03		135	153
OpenCUA-7B 2026.03		216	235
Computer-Use-Preview 2026.03		361	296
Agent-S3 w/ GPT-5 2026.03		623	628
CoAct-1 2026.03		924	857