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

Verification of Digital Agent Trajectories on VisualWebArena (VWA) and OSWorld

86Accuracy

GPT-5

53.7662.1370.578.87Jul 15, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.07
86858721
2025.07
84868262
2025.07
828978106
2025.07
818778106
2025.07
81867884
2025.07
808876128
2025.07
788373137
2025.07
788871117
2025.07
7791711512
2025.07
7688711411
2025.07
7689701512
2025.07
768273106
2025.07
7492642118
2025.07
7492651714
2025.07
7490641915
2025.07
7284651611
2025.07
7080622216
2025.07
6994552723
2025.07
6992572521
2025.07
6894552724
2025.07
6692532724
2025.07
6590512925
2025.07
6593512825
2025.07
6589542522
2025.07
6196423634
2025.07
6096403735
2025.07
6092443329
2025.07
5596344139