Local and cloud model support
Jan lets you choose from open models or connect to online providers from the same app, which makes it easier to switch between local and hosted AI workflows.
Jan
RevendiquerJan is an open-source AI assistant that can run local models on your own hardware or connect to cloud models like GPT and Claude. It also includes a local OpenAI-compatible API server, agent workflows, and integrations for private development and everyday chat.
Overview
Jan is an open-source AI assistant and ChatGPT alternative designed to run on your own hardware or connect to cloud models when you want them. The product positions itself as a local, privacy-first option, while still supporting popular model providers such as GPT, Claude, Gemini, Llama, Mistral, Qwen, DeepSeek, Gemma, and Kimi.
Beyond chat, Jan is presented as a broader AI workspace with built-in models, autonomous agents, a local OpenAI-compatible API server, and integration options for developer tools and MCP servers. The core appeal is control: you can keep data on your machine, switch between local and frontier models, and use Jan in desktop, web, terminal, or server-based workflows.
Core capabilities
Context memory
The product page highlights memory in progress, with context and preferences carried across sessions so users do not need to repeat themselves.
Built-in foundation models
Jan ships with foundation models designed for reasoning, agents, and deep research, and the docs say they are ready to run on your computer.
Local API server
Jan includes a local OpenAI-compatible API server powered by llama.cpp, so developers can build private or offline-capable apps against a familiar endpoint.
Agent workflows
The docs describe autonomous agents that can read files, manage calendars, and take actions through tools like WhatsApp, Discord, or Slack.
Integration ecosystem
Jan supports integrations with Claude Code, OpenClaw, and MCP servers, extending it beyond a standalone chat app.
Ways teams and individuals use Jan
Private everyday chat
Use Jan as a desktop chat app when you want a local-first assistant that keeps data on your own machine and can still fall back to cloud models when needed.
Offline-capable app development
Use the local API server to point internal tools or prototypes at an OpenAI-compatible endpoint without sending requests to a third-party cloud service.
Long-running research tasks
Use Jan’s agent features to hand off research, report drafting, or analysis tasks that can run for an extended period on your hardware.
Model evaluation and switching
Use the model switcher to compare local and hosted models in one place when you need flexibility for speed, quality, or data-handling reasons.
Tool-connected workflows
Use the integration ecosystem to connect Jan with Claude Code, MCP servers, or messaging and collaboration tools in a larger workflow.
Pros and Cons
Pros
- Runs locally on your hardware for private, offline-capable use.
- Supports both local models and several well-known cloud model providers.
- Includes an OpenAI-compatible local API server for application development.
- Offers autonomous agent workflows and integration hooks beyond basic chat.
- Open source and available for self-hosting.
Cons
- The source pages do not publish pricing tiers, so the commercial model is unclear from the reviewed pages.
- Some product areas, such as memory, are still marked as coming soon on the homepage.
- The evidence set only partially covers integrations and workflow details, so the full feature set is not completely documented here.
FAQ
What platforms does Jan support?
Jan runs on macOS, Windows, and Linux, and the docs also describe a web app, a local server, and a CLI. The desktop app can be installed without creating an account.
Can Jan use both local and cloud models?
Jan can run open-source models locally on your own hardware, or connect to cloud models such as GPT, Claude, Gemini, Llama, Mistral, Qwen, DeepSeek, Gemma, and Kimi as shown on the site.
Is Jan open source?
The product site describes Jan as open source, and the docs state that it is Apache 2.0 licensed and built in public. The handbook also says anyone can use it, contribute to it, or host it themselves.
Does Jan have an API server?
Jan provides a built-in OpenAI-compatible local API server. The docs say it runs entirely on your computer and can be used as a drop-in replacement for cloud APIs when building private, offline-capable applications.
How much does Jan cost?
The source indicates free and open-source availability, but it does not provide pricing tiers or subscription details on the pages reviewed.
Quick Facts
- Category
- AI Chat / AI Assistant
- Platforms
- macOS, Windows, Linux, web, terminal, server
- Source domain
- jan.ai
- License
- Apache 2.0
- Primary use
- Local and cloud AI chat, agents, and private development workflows
- Pricing
- Free and open source; no published tier details on reviewed pages