OpenAI Blog company blog

SWE-bench Verified no longer measures frontier coding capabilities

SWE-bench Verified is increasingly contaminated and mismeasures frontier coding progress. Our analysis shows flawed tests and training leakage. We recommend SWE-bench Pro.
Points
343
HN comments
181
Submissions
1
Cited by
7 commenters

AI & Machine LearningSoftware Engineering2026

On Hacker News

SubmittedTitle on HNPointsComments
Apr 26, 2026SWE-bench Verified no longer measures frontier coding capabilities by kmdupree343181

Cited in comments

Linked from 10 Hacker News comments by 7 different people, from Mar 2026 to Jul 2026.

Most recent:

Search every comment that links here →

More from OpenAI Blog

All 235 posts →

Related posts from the same era