Kimi K2.7 Code

Kimi K2.7 Code is Moonshot AI's coding-focused model, aimed at people who want an open-weight agent model that also reads images. The Hugging Face card gives about 1T total parameters with 32B active per token, a 256K context, and a Modified MIT licence. It builds on Kimi K2.6 and the card says it uses about 30% fewer thinking tokens. OpenRouter lists it from 12 June 2026 with a 262,144-token window, text and image input, at $0.60 per million input tokens and $3.50 per million output tokens. The catch is that thinking mode is forced on and cannot be turned off, and instant mode is not supported, so short edits cost more than they would on a non-reasoning model. Video input is experimental and only works through the official API. The card gives no release date of its own. Figures come from the Hugging Face card and the OpenRouter model list. We have not run it ourselves. We would use it for longer coding tasks, not quick completions.