Leaderboards

KimiKimi K3

moonshotai/kimi-k3
Evaluation report
Overall score
71.41
95% CI [70.46, 72.39]
Based on 285 sessions
Rank 11 of 11 models on the Overall leaderboard
Three goals
● Worlds 50%62.31
● Stories 25%68.24
● Preferences 25%92.80
Six dimensions
414799598693
BasicsLogicKnowledgeDiversityContent LogicInteraction
BasicsLogicKnowledgeDiversityContent LogicInteraction40

Scores by language

Tested November 1, 2026
LanguageRankOverall95% confidence intervalWorldsStoriesPreferencesSessions
Overall11 / 1171.4170.5 – 72.462.368.292.8285
Chinese10 / 1170.3169.1 – 71.561.267.591.3150
English11 / 1172.6471.3 – 74.063.569.194.4135