Moonshot: kimi-k2-turbo-preview API
A high-speed preview version of the Kimi K2 series by Moonshot AI. Built on a 1-trillion parameter Mixture-of-Experts (MoE) architecture, it is specifically optimized for agentic tasks, coding, and real-time interactions. It maintains superior instruction-following while significantly increasing inference speed (up to 60-100 tokens/s), making it an ideal choice for long-context handling and complex tool-calling workflows.
- Context window: 262,144 tokens
- Max output: 262,144 tokens
- Input: text, image, file
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- File input: Supported
- Released: 2025-09-05
- Knowledge cutoff: 2025-10-14
Frequently Asked Questions
What is the context window of kimi-k2-turbo-preview?
kimi-k2-turbo-preview supports a context window of up to 262,144 tokens.
Does kimi-k2-turbo-preview support function calling?
Yes. kimi-k2-turbo-preview supports tool / function calling.
Does kimi-k2-turbo-preview support reasoning?
Yes. kimi-k2-turbo-preview is a reasoning-capable model.
What is the knowledge cutoff of kimi-k2-turbo-preview?
The knowledge cutoff of kimi-k2-turbo-preview is 2025-10-14.