Announcement_18
🎉 Our framework reached 81.2% on SWE-bench Verified (as of Mar 27, 2026) by reusing experience mined from past agent trajectories (STAIR), higher than every entry on the official leaderboard 🚀
🎉 Our framework reached 81.2% on SWE-bench Verified (as of Mar 27, 2026) by reusing experience mined from past agent trajectories (STAIR), higher than every entry on the official leaderboard 🚀