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

GUI Agent Task Success on AndroidWorld (online)

48.7Task Success Rate

GPT-5

1.69213.89626.138.304Feb 25, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2026.02
48.7
2026.02
42.6
2026.02
42.6
2026.02
42.6
2026.02
40.9
2026.02
39.1
2026.02
37.4
2026.02
36.5
2026.02
34.8
2026.02
32.2
2026.02
30.6
2026.02
30.4
2026.02
29.6
2026.02
29.6
2026.02
27.2
2026.02
27
2026.02
25.2
2026.02
23.7
2026.02
20.9
2026.02
18.3
2026.02
16.5
2026.02
15.7
2026.02
7.8
2026.02
7
2026.02
3.5