Moonshot AI

Moonshot AI's native multimodal model delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm. Built on Kimi K2 with continued pretraining over approximately 15T mixed visual and text tokens.

Moonshot AI's most advanced open reasoning model. Extends the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter MoE architecture with 32B active parameters.

September update of Kimi K2. Large-scale Mixture-of-Experts (MoE) language model with 1 trillion total parameters and 32B active per forward pass. Supports long-context inference up to 256K.

1-trillion-parameter MoE agentic model excelling at tool use and software engineering. Open-sourced under MIT.

Moonshot's first long-chain-of-thought multimodal reasoning model, matching o1 on math and coding at launch.