datacurve-ai/deep-swe
Measuring frontier coding agents on original, long-horizon engineering tasks
3 best points · 0 comments · 2 Hacker News posts · 1,893 GitHub stars · View on GitHub
Hacker News discussions
- DeepSWE Audit: DeepSeek-v4-pro results are unreliable (3 points, 0 comments, by eunos, 2026-06-04)
- DeepSWE results are unreliable – 3/3 DSv4 "failed" tasks solved with same model (3 points, 0 comments, by theanonymousone, 2026-06-04)