What is Jan?
Jan is an open-source ChatGPT replacement that lets users run open-source AI models locally or connect to cloud models such as GPT, Claude, Gemini, Llama, Mistral, Qwen, DeepSeek, Gemma and Kimi. Jan is built in public on GitHub, where it has 45K stars, and has passed 6 million downloads.
Key Features
- Local model execution: Jan runs open-source AI models directly on the user's machine, giving users control over where their model workloads run.
- Cloud model connections: Users can plug in online models from providers including OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini and Gemma), Meta (Llama), Mistral AI, Alibaba (Qwen), DeepSeek and Moonshot AI (Kimi).
- Model choice in one interface: Jan combines open models and online models in a single application, so users can switch between them rather than juggling separate clients.
- Memory (coming soon): Jan is slated to carry context over between sessions so users do not have to repeat themselves; the feature is described as remembering context and preferences.
- Open-source codebase: Jan is developed in public on GitHub, and the project reports 45K stars, allowing developers to inspect and follow the codebase.
- HuggingFace model hub presence: The project lists 123 models on HuggingFace, giving users a catalog of open models to draw from.
- Community channels: Jan maintains a Discord community and a GitHub repository as its primary support and development surfaces.
- Mac availability: Jan offers a download for Mac, and the project reports over 7 million downloads.
Use Cases
- Privacy-conscious developers: A developer who wants to run open-source models on their own machine rather than sending prompts to a hosted service can use Jan for local model execution.
- Multi-provider workflows: A user who works across GPT, Claude, Gemini and open models can consolidate those connections in Jan instead of maintaining separate accounts and clients.
- Open-source model experimentation: Developers exploring models from Meta, Mistral AI, Alibaba, DeepSeek, Google and Moonshot AI can access them alongside cloud options in one place.
- Context-heavy personal use: Once the coming-soon memory feature ships, users with recurring preferences and ongoing projects, such as design or prototyping work, are the intended beneficiaries of carried-over context.
- Community-driven development: Developers who want to follow or contribute to an openly built AI client can engage through Jan's GitHub repository and Discord community.
Who It Is For
Jan is aimed at developers and technically comfortable users who want an open-source alternative to ChatGPT and want control over which models they use, whether open models run locally or cloud models from major providers. It suits people who prefer to inspect the code behind their tools, follow development in public on GitHub, and consolidate model access in a single application rather than committing to one vendor's hosted chat product.
Pricing
Jan is free and open source, as stated on its website. No paid tiers or pricing figures are listed in the source material.
Frequently Asked Questions
What is Jan used for?
Jan is an open-source ChatGPT replacement used to run open-source AI models locally or to connect to cloud models such as GPT, Claude, Gemini, Llama, Mistral, Qwen, DeepSeek, Gemma and Kimi from a single application.
Is Jan free?
Yes. Jan is described on its website as free and open source, and its code is developed in public on GitHub.
What models does Jan support?
Jan supports open models as well as online models from OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini and Gemma), Meta (Llama), Mistral AI, Alibaba (Qwen), DeepSeek and Moonshot AI (Kimi). The project also lists 123 models on HuggingFace.
Does Jan run models locally?
Yes. Jan can run open-source AI models locally on the user's machine, and it can also connect to cloud models when users prefer hosted options.
Does Jan have a memory feature?
Jan lists a memory feature as coming soon. According to the project, it will carry context over so users do not have to repeat themselves, remembering context and preferences across sessions.
