Claude Code, Codex, and Pi show similar accuracy on SWE-Bench Pro with about double the cost for Claude Code and Codex compared to Pi
Read the original at aistack.imec-int.com→Original headline: "Benchmarking Claude Code, Codex and Pi on SWE-Bench Pro: Same Accuracy, 2x Cost"
Coverage timeline
- Sep 10, 06:50 UTC Hacker News (AI) lead source Benchmarking Claude Code, Codex and Pi on SWE-Bench Pro: Same Accuracy, 2x Cost