DeepSWE

A new standard for agentic coding benchmarks.

A benchmark for evaluating agentic, long-horizon software-engineering workflows using repository search, multi-file edits, and verification tasks.

Recent stories

1 linked story

releasePRIMARY2026-05-27

DeepSWE benchmarks GPT-5.5 at 70% on 113 tasks across 91 repos

DeepSWE launched a coding benchmark built from 113 original tasks across 91 repos and five languages, with GPT-5.5 leading at 70%. The setup is meant to better reflect repo search, multi-file edits, and verification in real agent workflows.