Moonshot: kimi-k2-thinking-turbo API

A high-speed reasoning model in the Kimi K2 series by Moonshot AI. As the optimized version of kimi-k2-thinking, it is built on a 1-trillion parameter MoE architecture and natively supports explicit Chain-of-Thought (CoT). By leveraging native INT4 Quantization-Aware Training (QAT), it boosts inference speed from 8 tokens/s to 50-60 tokens/s (peaking at 100 tokens/s), making it ideal for high-frequency interactive long-horizon agentic workflows.

Frequently Asked Questions

What is the context window of kimi-k2-thinking-turbo?

kimi-k2-thinking-turbo supports a context window of up to 262,144 tokens.

Does kimi-k2-thinking-turbo support function calling?

Yes. kimi-k2-thinking-turbo supports tool / function calling.

Does kimi-k2-thinking-turbo support reasoning?

Yes. kimi-k2-thinking-turbo is a reasoning-capable model.

What is the knowledge cutoff of kimi-k2-thinking-turbo?

The knowledge cutoff of kimi-k2-thinking-turbo is 2024-07-31.