AIHacker News38 minutes ago
Benzi Harness Beats Claude Code on SWE-bench
1 min read
Benzi's code intelligence harness resolved 78.2% of 500 real GitHub issues on SWE-bench Verified at under 10 cents per fix. A comparison across 24 issues in 10 languages shows Benzi's cost and lines-read slope are lower than Claude Code and CodeGraph as difficulty increases.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
Summary by ByteBrief
