Coding · benchmark

SWE-bench Verified

Fix real GitHub issues and check patches with repository tests.

What it measures

Resolved tasks (%)

Docker environments; agent system and model-only tracks differ.

Public evaluation code. Check upstream code and dataset terms.

Open source / dataset · Run instructions · Official results

Reported ranking tracks

• SWE-bench Verified · agent systems · 180 entries · captured 2026-10-07T07:01:28Z
• SWE-bench Verified · mini-SWE-agent · 47 entries · captured 2026-10-07T07:01:28Z

Explore all evaluations · Find a competition