Moonshot: kimi-k2-thinking API
The first large language model from Moonshot AI that natively supports Chain-of-Thought (CoT) reasoning. Built on a 1-trillion parameter MoE architecture (32B active), it is trained using large-scale Reinforcement Learning (RL) to "think before acting." Its breakthrough lies in stable long-horizon agency, capable of executing 200–300 sequential tool calls (search, coding, data analysis) without drift, making it a premier model for complex reasoning and autonomous workflows.
- Context window: 262,144 tokens
- Max output: 262,144 tokens
- Input: text
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- File input: Supported
- Released: 2025-11-06
- Knowledge cutoff: 2024-07-31
Frequently Asked Questions
What is the context window of kimi-k2-thinking?
kimi-k2-thinking supports a context window of up to 262,144 tokens.
Does kimi-k2-thinking support function calling?
Yes. kimi-k2-thinking supports tool / function calling.
Does kimi-k2-thinking support reasoning?
Yes. kimi-k2-thinking is a reasoning-capable model.
What is the knowledge cutoff of kimi-k2-thinking?
The knowledge cutoff of kimi-k2-thinking is 2024-07-31.