Kimi K2 Thinking API
About
Build advanced reasoning agents with Kimi K2 Thinking, a specialized AI model optimized for complex logic and chain-of-thought tasks. This model delivers available completely free of charge, native tool calling support, open weights architecture. Access Kimi K2 Thinking via the Kimi For Coding API with up to 33K output tokens.
Capabilities
Input Modalities
text
Output Modalities
text
Reasoning
Yes
Structured Output
Yes
Tool Use
Yes
WeightsOpen
Limits
Context Window
262K
Tokens
Input Limit
262K
Tokens
Max Output
33K
Tokens
Pricing
Standard (per 1M tokens)
Input
Free
Output
Free
Caching (per 1M tokens)
Read
Free
Write
Free
Frequently Asked Questions
Kimi K2 Thinking by Kimi For Coding costs Free per 1M input tokens and Free per 1M output tokens. Cached reads cost Free per 1M tokens. Cache writes cost Free per 1M tokens.
Model Details
ID
kimi-k2-thinking
Provider
Kimi For CodingFamily
kimi-thinking
Release Date
Nov 1, 2025
Knowledge Cutoff
Jul 1, 2025
API Integration
NPM Package
@ai-sdk/anthropic
Environment Variables
KIMI_API_KEY
API Base URL
https://api.kimi.com/coding/v1