Benchmarks
Only 3 of 16 products have their own SWE-bench entry
Jules (52.2% Verified, Dec 2024), Factory Code Droid (37.0%, Jun 2024) and OpenHands (71.8% with GPT-5) are on the official leaderboard. Devin, Claude Code, Cursor, Copilot and every app builder have no product entry. Codex appears only as a model inside another harness.