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

GUI Agent Interaction on OSWorld (Step Success Rate)

42.9Success Rate (Max Steps: 15)

Claude 4.5 Sonnet

1.40412.17722.9533.723May 28, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.05
42.9-58.1-
2026.05
42.833.847.4-
2026.05
33.124.535.2-
2026.05
32.122.141.6-
2026.05
31.9-40-
2026.05
31.726.336.3-
2026.05
29.715.534.1-
2026.05
29.221.633.9-
2026.05
28.121.141-
2026.05
27.114.729.4-
2026.05
26-31.3-
2026.05
24.59.527.3-
2026.05
24.312.528.2-
2026.05
241125.8-
2026.05
4.41.15-
2026.05
30.73.9-
2026.06
---6.8
2026.06
---8.2
2026.06
---9
2026.06
---9.3
2026.06
---9.5
2026.06
---10.3
2026.06
---12
2026.06
---12.8
2026.06
---8.9
2026.06
---10.6
2026.06
---14.2
2026.06
---16.3
2026.06
---17
2026.06
---17.4
2026.06
---17.1