API
Agents/Kimi K2.5

Kimi K2.5

by Moonshot AI

best score
70.8%
Best Score
bash-only
Best Leaderboard
1
Models Used
Yes
Open Source
EntryScore
Kimi K2.5 (high reasoning)70.8%
mini-SWE-agent + Kimi K2.5 (high reasoning)70.8%
Kimi K2.567.3%
Kimi K2 Thinking63.4%
Kimi K2 Instruct43.8%