List your company

Software Engineering Issue Resolution on SWE-bench 500 (test)

72.2Resolution Scoreagyn
64.71266.65668.670.544Feb 1, 2026

Evaluation results

MethodLinks
agynModel=GPT-5 / GPT-5-codex, Reasoning Level=medium reasoning2026.0272.2
OpenHandsModel=GPT-52026.0271.8
mini-SWE-agentModel=GPT-5.2, Reasoning Level=high reasoning2026.0271.8
mini-SWE-agentModel=GPT-5, Reasoning Level=medium reasoning2026.0265