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

Agentic Coding on SWE-bench 50 tasks sampled

12Resolved Count

Dense

0.563.536.59.47Jun 22, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
1224
2026.06
1224
2026.06
1224
2026.06
1224
2026.06
918
2026.06
918
2026.06
816
2026.06
816
2026.06
816
2026.06
816
2026.06
816
2026.06
714
2026.06
714
2026.06
612
2026.06
612
2026.06
612
2026.06
48
2026.06
12
2026.06
12
2026.06
12