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

Software Engineering on SWE-bench Verified (Pass@1 and Time Metrics)

72Pass@1

GPT-5 mini (2025-08-07)

22.80835.57948.3561.121May 6, 2026May 16, 2026May 26, 2026Jun 5, 2026Jun 15, 2026Jun 25, 2026Jul 6, 2026
Updated 18d ago

Evaluation Results

MethodLinks
2026.07
72------
2026.07
66.8------
2026.07
66.5------
2026.07
62.5------
2026.07
62------
2026.07
59.8------
2026.07
58.3------
2026.07
58------
2026.07
57.8------
2026.07
57.3------
2026.07
56------
2026.07
51.9------
2026.07
50.5------
2026.07
50------
2026.07
48------
2026.07
47.5------
2026.07
45.2------
2026.07
43.7------
2026.05
42.2172.9168.12902822,14639.8
2026.05
36.8268258.24103952,35330.9
2026.05
29.566.380.41702061,40536.7
2026.07
25.2------
2026.05
24.7133.1138.43003121,59725.9