kimi-k2-thinking · Moonshot AI
Kimi K2 Thinking is Moonshot AI's most advanced open-source inference model to date, extending the K2 series into intelligent agent and long-context inference domains. The model is built on the trillion-parameter mixture of experts (MoE) architecture introduced by Kimi K2, activating 32 billion parameters per forward pass and supporting a context window of 256,000 tokens.
Kimi K2 Thinking is Moonshot AI's most advanced open-source inference model to date, extending the K2 series into intelligent agent and long-context inference domains. The model is built on the trillion-parameter mixture of experts (MoE) architecture introduced by Kimi K2, activating 32 billion parameters per forward pass and supporting a context window of 256,000 tokens.
Kimi K2 Thinking has a 262,144 token context window.
On AIHubMix, Kimi K2 Thinking costs $0.548 per million input tokens and $2.192 per million output tokens. Cached input reads are billed at $0.137 per million tokens.
Kimi K2 Thinking accepts text input.
Kimi K2 Thinking supports thinking, tool calling, function calling and structured outputs. Per-protocol parameter support is listed in the capability table on this page.
Kimi K2 Thinking is available through the AIHubMix unified API. The API is OpenAI-compatible: point your OpenAI SDK at https://aihubmix.com/v1, use your AIHubMix API key, and set the model name to kimi-k2-thinking — no other code changes needed.
Kimi K2 Thinking is developed by Moonshot AI. AIHubMix aggregates it alongside models from other providers behind one API and one bill.
Kimi K3 is Kimi’s flagship model for long-horizon coding and end-to-end knowledge work…
coding-kimi-k3-free is the open and free version of coding-kimi-k3. To maintain reliable…
Kimi K2.7 Code is Kimi’s most intelligent Coding model, capable of completing programming…
High-Speed version of Kimi K2.7 Code model, with output speed of approximately 180…
Kimi K2.6 is Kimi's latest and most intelligent model, with stronger and more stable…
Use Kimi K2 Thinking via the AIHubMix unified API — one interface for every major LLM.