Open-weight agentic model
Kimi K2
Best for: Tool use, coding and agents
Very large mixture-of-experts model; not practical to run at home.
AppAPIOpen weights
🇨🇳 China · China
Moonshot AI (Kimi) models explained
Beijing-based Moonshot AI makes Kimi, an assistant known for long-context reading and research, and has released large open-weight models designed for agentic tasks.
All model families
Reasoning
Kimi thinking models
Best for: Deep research and hard problems
Reasoning versions used in the Kimi assistant.
AppAPI
Assistant
Kimi assistant
Best for: Long PDFs, comparisons, research
Web and mobile app.
App
Model names and versions change often. Check the official documentation for the newest releases.
⚡ Which model should you use?
- Long documents and research: kimi.com.
- Developers building agents: Kimi API.
Data, privacy and your region
Cloud services from Chinese companies process data under Chinese law and the provider’s privacy policy. Many organisations in the UK, EU and US restrict their use for sensitive data. The open-weight models can instead be run locally or on your own servers, where your data stays under your control.
Frequently asked questions
Can I run Kimi K2 locally?
The weights are public, but the model is extremely large and needs data-centre hardware; most people use the website, app or API.