Model lineup
The Kimi family as of October 2026 — compiled from official public pages. Defer to official releases.
| Model | Positioning | Context | List price (per M tokens) |
|---|---|---|---|
| kimi-k3.1-proLatest | Current flagship for complex tasks | 1,048,576 | $2.00 in / $10.00 out |
| kimi-k3 | Flagship: 2.8T parameters, natively multimodal | 1,048,576 (K3 series) | See platform |
| kimi-k2.7-code | Coding specialist | See platform | See platform |
| kimi-k2.6 | April 2026 flagship, complex problem solving | See platform | See platform |
kimi-k3.1-pro Latest · October 2026
The current flagship. Artificial Analysis Intelligence Index 55 — rank #4 of 215 models — at $2.00/$10.00 per million tokens (on par with the market median), with a 90% cache discount. Text + image input, text output, 1M-token context (≈1,500 A4 pages). Output speed 73.5 tokens/s, slightly below the 82 average — a depth-oriented profile.
kimi-k3 July 2026
The K3 generation flagship: 2.8 trillion parameters with a new architecture, natively multimodal, 1M-token context. Built for long-horizon coding, knowledge work and deep reasoning. In official blog evaluations it handled GPU kernel optimization at the level of Claude Fable 5 and built a working Triton-like GPU compiler (MiniTriton) from scratch.
kimi-k2.7-code
Coding specialist for software engineering: code generation, analysis, refactoring and long-horizon terminal tasks. Pair it with the K3-generation agent loop for screenshot-driven frontend and CAD work.
kimi-k2.6 April 2026
The April 2026 flagship: "complex problems, solved with ease" — code, analysis, spreadsheets and slides in one model. Still available on the platform for workflows pinned to the K2 generation.