GitHub launches ReviewBench open benchmark to evaluate AI code reviewers
GitHub has released ReviewBench, an open benchmark designed to evaluate whether AI code review agents identify critical issues without adding noise, based on 219 pull requests across 19 languages.