Kimi Code: Kimi K2.7-Code released
Open-weight (Modified MIT) 1T-parameter MoE coding model (32B active, 256K context) built for the Kimi Code CLI; ~30% fewer reasoning tokens than K2.6.
Open-weight (Modified MIT) 1T-parameter MoE coding model (32B active, 256K context) built for the Kimi Code CLI; ~30% fewer reasoning tokens than K2.6.