# Old Jan Homepage

Turns your computer or server into an self-hosted ChatGPT alternative, complete with a OpenAI-equivalent API.
- Runs Local AI or connects to Remote APIs
- Download and run popular open-source models, fully offline
- Local API Server

Software that you own
- [ ] Take inspiration from Twenty's writeup: https://twenty.com/story.
- Jan is AGPLv3 licensed. You own, not rent the software
- You get the simplicity of SaaS, with the freedom and cost savings of Open Source
- Features
- Thriving Community
- Welcoming community of hundreds of developers that share expertise and support
- Open Roadmap
- Missing a feature? Start a discussion, report an issue, contribute code or fork the repo
- Secure by Design
- Benefit from the collective oversight of a global community that quickly identifies and resolves issues

Jan is an open-source, self-hosted alternative to OpenAI's platform. It allows you to build, run, and fine-tune (coming soon) on your own desktop or server. Featuring a local API server that mirrors OpenAI's, Jan operates directly from your device (localhost:1337), enabling immediate access to AI capabilities with complete control over your data.

To use Jan, please refer to the quick start guide [here](https://jan.ai/docs/quickstart).

## Why Jan?

We believe in the need for an open source AI ecosystem.

We're focused on building infra, tooling and [custom models](https://huggingface.co/janhq) to allow open source AIs to compete on a level playing field with proprietary offerings.

Read more about our mission and culture [here](/about).

### 💻 Self-hosted AI

Jan runs 100% on your own machine, predictably, privately and offline. No one else can see your conversations, not even us.

### 🏗️ Extensions

Jan ships with a local-first, AI-native, and cross platform [extensions framework](/developer/build-extension). Developers can extend and customize everything from functionality to UI to branding. In fact, Jan's current main features are actually built as extensions on top of this framework.

### 🗂️ Open File Formats

Jan stores data in your [local filesystem](/developer/file-based). Your data never leaves your computer. You are free to delete, export, migrate your data, even to a different platform.

### 🌍 Open Source

Both Jan and [Nitro](https://nitro.jan.ai), our lightweight inference engine, are licensed via the open source [AGPLv3 license](https://github.com/janhq/jan/blob/main/LICENSE).

## Features

# Legacy
- Compatible with [open-source models](/guides/using-models) (GGUF via [llama.cpp](https://github.com/ggerganov/llama.cpp), TensorRT via [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), and [remote APIs](https://platform.openai.com/docs/api-reference))
- Compatible with most OSes: [Windows](/install/windows/), [Mac](/install/mac), [Linux](/install/linux), with GPU acceleration through [llama.cpp](https://github.com/ggerganov/llama.cpp)
- Stores data in [open file formats](/developer/file-based)
- Local API [server mode](/guides/using-server)
- Customizable via [extensions](/developer/build-extension)
- And more in the [roadmap](https://github.com/orgs/janhq/projects/5/views/16). Join us on [Discord](https://discord.gg/5rQ2zTv3be) and tell us what you want to see!

## Design Principles

- [ ] Website's Principles Section

## Architecture

## Roadmap

## Quick Guides

- **[Models](/docs/models):** Manage and deploy AI models effectively.
- **[Threads](/docs/threads):** Handle multiple operations and tasks with ease.
- **[Local Server](/docs/local-api):** Utilize Jan's OpenAI-Equivalent API for your applications.
- **[Advanced Settings](/docs/settings):** Navigate through Jan's advanced settings for optimal customization.
- **[Extensions](/docs/extensions):** Expand Jan's capabilities with multi-inferences and specific settings for each extension.
- **[Integrations](/integrations):** Integrate Jan with key tools and platforms to enhance your development environment.

## Legacy: About Page

Jan is a [open-source](https://en.wikipedia.org/wiki/Open_source), [local-first](https://www.inkandswitch.com/local-first/) tool to [create, customize and use AI](https://www.gatesnotes.com/AI-agents) for everyday tasks.

You can:
- Run locally using [open-source LLMs](https://huggingface.co/models?pipeline_tag=text-generation) or connect to cloud AIs like [ChatGPT](https://openai.com/blog/openai-api) or [Google](https://ai.google.dev/)
- Fine-tune AI with specific knowledge
- Search the web and other databases
- Connect AI to your everyday tools and (with your permission) do work on your behalf

Longer-term, Jan is building a cognitive framework for future robots. We envision a world where we have personal or company robots that we continually improve and customize, growing together with us.


### Why do we exist

At Jan, our mission is to advance human-machine collaboration. We achieve this through delivering the best open-source, local-first tools to allow users to run, customize and tinker with AI.

### What's different about it?

| | Status Quo | Jan |
| ---------------------------------------------------------- | -------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- |
| Ownership | AI Monopolies owned by Big Tech | AI that you own and control |
| Openness? | Closed-source | [Open-source (AGPLv3)](https://github.com/janhq/jan/blob/main/LICENSE) |
| Your role | Consume | Create, Tinker and Customize |
| Approach | Cloud | [Local-first](https://www.inkandswitch.com/local-first/), running 100% on your computer |
| Data | Data stored on their servers | Data stored in your local filesystem in open, non-proprietary file formats |
| Privacy | Privacy? | Runs 100% on your own machine, predictably, privately and offline |
| Transparency | "Black Box" | Runs predictability with code available to tinker and customize |
| What happens if there's an outage or goes out of business? | Your life's work held hostage in the cloud in proprietary data formats[^1] | Continues to run 100% on your computer, your data is safe in your local folder |
| Driving Philosophy | Monetize your users | [Privacy as a human right](https://en.wikipedia.org/wiki/Right_to_privacy) and the [Right to Repair](https://www.repair.org/) |

### How do I get it?

You can install and start using Jan in less than 5 minutes, from [jan.ai](https://jan.ai) or our [Github repo](https://github.com/janhq/jan).

You can read the [User Guide](/docs/user-guide) if you need some help to get started.

### What license is the code under?

Jan is licensed under the [AGPLv3 License](https://github.com/janhq/jan/blob/main/LICENSE).

We happily accept pull requests, however we do ask that you sign a [Contributor License Agreement](https://en.wikipedia.org/wiki/Contributor_License_Agreement) so that we have the right to relicense your contributions[^2].

We also have a [Contributor Program](/docs/team/contributor-program) to provide ownership and upside to contributors who have made significant contributions to the project.

### What was it built with?

Jan is pragmatically built using Javascript at the application level, and C++ at the Inference level (which we have refactored into [Nitro](https://nitro.jan.ai)). We use [Electron](https://www.electronjs.org/) to build the desktop client.

We have made similar choices to the [Next.js Enterprise Javascript Stack](https://vercel.com/templates/next.js/nextjs-enterprise-boilerplate), which is a [battle-tested](https://nextjs.org/showcase/enterprise) framework for building enterprise-grade applications.

### Contact

Drop us a message in our [Discord](https://discord.gg/af6SaTdzpx) and we'll get back to you.

- `#general`: for general discussion
- `#get-help`: for bug reports and troubleshooting
- `#roadmap`: for feature requests and ideas

### Footnotes

[^1]: Credit to Obsidian's original website
[^2]: Credit to [Discourse's About Page](https://www.discourse.org/about)

## Legacy: About Page

he following low-level docs are aimed at core contributors.

We cover how to contribute to the core framework (aka the `Core SDK`).

:::tip
If you are interested to **build on top of the framework**, like creating assistants or adding app level extensions, please refer to [developer docs](/developer) instead.
:::

### Jan Framework

At its core, Jan is a **cross-platform, local-first and AI native framework** that can be used to build anything.

#### Extensions

Ultimately, we aim for a `VSCode` or `Obsidian` like SDK that allows **devs to build and customize complex and ethical AI applications for any use case**, in less than 30 minutes.

In fact, the current Jan [Desktop Client](https://jan.ai/) is actually just a specific set of extensions & integrations built on top of this framework.

:::tip
We encourage devs to fork, customize, and open source your improvements for the greater community.
:::

#### Cross Platform

Jan follows [Clean Architecture](https://blog.cleancoder.com/uncle-bob/2012/08/13/the-clean-architecture.html) to the best of our ability. Though leaky abstractions remain (we're a fast moving, open source codebase), we do our best to build an SDK that allows devs to **build once, deploy everywhere.**

**Supported Runtimes:**

- `Node Native Runtime`, good for server side apps
- `Electron Chromium`, good for Desktop Native apps
- `Capacitor`, good for Mobile apps (planned, not built yet)
- `Python Runtime`, good for MLOps workflows (planned, not built yet)

**Supported OS & Architectures:**

- Mac Intel & Silicon
- Windows
- Linux (through AppImage)
- Nvidia GPUs
- AMD ROCm (coming soon)

Read more:

- [Code Entrypoint](https://github.com/janhq/jan/tree/main/core)
- [Dependency Inversion](https://en.wikipedia.org/wiki/Dependency_inversion_principle)

#### Local First

Jan's data persistence happens on the user's local filesystem.

We implemented abstractions on top of `fs` and other core modules in an opinionated way, s.t. user data is saved in a folder-based framework that lets users easily package, export, and manage their data.

Future endeavors on this front include cross device syncing, multi user experience, and more.

Long term, we want to integrate with folks working on [CRDTs](https://www.inkandswitch.com/local-first/), e.g. [Socket Runtime](https://www.theregister.com/2023/04/11/socket_runtime/) to deeply enable local-first software.

Read more:

- [Folder-based wrappers entrypoint](https://github.com/janhq/jan/blob/main/core/src/fs.ts)
- [Piping Node modules across infrastructures](https://github.com/janhq/jan/tree/main/core/src/node)

:::caution
Our local first approach at the moment needs a lot of work. Please don't hesitate to refactor as you make your way through the codebase.
:::

#### AI Native

We believe all software applications can be natively supercharged with AI primitives and embedded AI servers.

Including:

- OpenAI Compatible AI [types](https://github.com/janhq/jan/tree/main/core/src/types) and [core extensions](https://github.com/janhq/jan/tree/main/core/src/extensions) to support common functionality like making an inference call.
- Multiple inference engines through [extensions, integrations & wrappers](https://github.com/janhq/jan/tree/main/extensions/inference-nitro-extension) _On this, we'd like to appreciate the folks at [llamacpp](https://github.com/ggerganov/llama.cpp) and [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM) for. To which we'll continue to make commits & fixes back upstream._

- [Code Entrypoint](https://github.com/janhq/jan/tree/main/core/src/api)

##### Fun Project Ideas

Beyond the current Jan client and UX, the Core SDK can be used to build many other AI-powered and privacy preserving applications.

- `Game engine`: For AI enabled character games, procedural generation games
- `Health app`: For a personal healthcare app that improves habits
- Got ideas? Make a PR into this docs page!

If you are interested to tackle these issues, or have suggestions for integrations and other OSS tools we can use, please hit us up in [Discord](https://discord.gg/5rQ2zTv3be).

:::caution
Our open source license is copy left, which means we encourage forks to stay open source, and allow core contributors to merge things upstream.
:::

## Legacy: About Page

The following docs are aimed at end users who want to troubleshoot or learn how to use the **Jan Desktop** application better.

:::tip
If you are interested to build extensions, please refer to [developer docs](/developer) instead (WIP).

If you are interested to contribute to the underlying framework, please refer to [framework docs](/docs) instead.
:::

### Jan Desktop

The desktop client is a ChatGPT alternative that runs on your own computer, with a [local API server](/guides/using-server).

### Features

- Compatible with [open-source models](/guides/using-models) (GGUF via [llama.cpp](https://github.com/ggerganov/llama.cpp), TensorRT via [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), and [remote APIs](https://platform.openai.com/docs/api-reference))
- Compatible with most OSes: [Windows](/install/windows/), [Mac](/install/mac), [Linux](/install/linux), with GPU acceleration through [llama.cpp](https://github.com/ggerganov/llama.cpp)
- Stores data in [open file formats](/developer/file-based)
- Local API [server mode](/guides/using-server)
- Customizable via [extensions](/developer/build-extension)
- And more in the [roadmap](https://github.com/orgs/janhq/projects/5/views/16). Join us on [Discord](https://discord.gg/5rQ2zTv3be) and tell us what you want to see!

### Models
Jan lets you run and manage different AI models on your own device. Choose between running AI models locally for privacy, like `Llama` or `Mistral`, or connecting to remote APIs, like `ChatGPT` or `Claude`. This feature supports customizing and optimizing AI to meet the needs of your projects directly, with a focus on maintaining speed and privacy. You can download and import models directly through Jan Hub, making it easy to expand your AI capabilities.

### Threads
Jan offers a simple and secure way to manage your AI interaction threads directly on your device. Every interaction with AI through Jan is saved as a thread, allowing you to keep a record of your exchanges. Jan provides tools to sort, remove, or review these threads efficiently. This ensures your AI interaction history is easily accessible, kept private, and well-managed.

### Local API Server
Jan’s local API server acts as a `drop-in replacement for OpenAI's API`, allowing you to run AI models on your own server. This setup offers AI functionality on your local machine, giving you complete control over your data and privacy.

### Inferences
Jan supports multiple inferences, including `Llama.cpp`, `TensorRT-LLM`, `Mistral API`, `OpenAI API`, `Groq API`, `LM Studio`, `Ollama`, and other `OAI-compatible servers`.

### Extensions
Jan’s architecture supports a variety of extensions, enabling enhanced functionality and inference options. Each extension is designed to meet specific needs, ensuring that Jan can be tailored to the unique demands of your projects.

### Integrations
Jan integrates with an array of tools and platforms to streamline your workflow. This includes seamless connections with services like `Discord`, `Open Interpreter`, `Raycast`, and `OpenRouter`.

### Assistants (Coming soon)
Jan will soon enable the creation of AI assistants for task automation and response generation. These assistants will be customizable, allowing for various interactions, from simple commands to complex conversations.

## Old About Page

### Open Source

We believe that the future of AI should be built on open-source foundations. Jan and Cortex are copy-left (AGPLv3).

We build on the shoulders of open source giants: [llama.cpp](https://jan.ai/docs/local-inference/llamacpp), [tensorrt-llm](https://jan.ai/docs/local-inference/tensorrtllm).

### Local & Private

By default, Jan runs AI locally on your device. Your data, models, and training dataset are stored on your local filesystem. No data leaves your computer. You own the underlying data, models & infrastructure.

### Build in Public

We openly share our roadmap, metrics, and strategy.

- [Jan's Kanban](https://github.com/orgs/janhq/projects/5)
- [Jan's Roadmap](https://github.com/orgs/janhq/projects/5/views/29)
- [Engineering Discussions (Discord)](https://discord.gg/AAGQNpJQtH)

### Sustainable

Jan is currently a bootstrapped startup. We balance technical invention with the search for a sustainable business model, by creating real value for our community.

We welcome business inquiries: [e-mail]

## Comparisons

> "I do not fear computers. I fear the lack of them." - Isaac Asimov

Jan was founded on the belief that AI should coexist with humans, not replace them.

We are committed to creating open, local-first products that give back human control, reject dark patterns & ecosystem lock-ins, while providing a comprehensive solution for advanced applications.

| | Alternatives | Jan |
| ----------------- | ---------------------------------- | --------------------------------------------- |
| Privacy | Questionable | Private, with enhanced data protection |
| Transparency | Closed source | Open source, and customizable |
| Ownership | Give away your IP | You own the AI (the IP) |
| Standardization | Custom APIs | 100% OpenAI Compatible |
| Customization | Infrastructure lock-in | Plug & play |
| Hosting | Monthly cloud bills | Self-hosted |
| Data | Cloud-based | Local filesystem based, with database options |
| Outage Resilience | Potential hostage & API throttling | Works offline |

Read more on our vision [here](https://jan.ai/about/vision).

|                                                            | Thinking Machines                                                                                                     | Status Quo                                                                 |
| ---------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------- |
| Ownership                                                  | Give users ownership and control                                                                                                  | AI Monopolies owned by Big Tech                                            |
| Openness?                                                  | Open Source [(Copyleft)](https://en.wikipedia.org/wiki/Copyleft#:~:text=Copyleft%20is%20the%20legal%20technique,be%20preserved%20in%20derivative%20works.)                                                        | Closed-source                                                              |
| Your role                                                  | Encourage users to Create, Tinker and Customize                                                                                                  | Consume                                                                    |
| Approach                                                   | [Local-first](https://www.inkandswitch.com/local-first/), running 100% on your computer                                       | Cloud                                                                      |
| Data                                                       | Data stored in your local filesystem in open, non-proprietary file formats                                                    | Data stored on their servers                                               |
| Privacy                                                    | Runs 100% on your own machine, predictably, privately and offline                                                             | Privacy?                                                                   |
| Transparency                                               | Runs predictability with code available to tinker and customize                                                               | "Black Box"                                                                |
| What happens if there's an outage or goes out of business? | Continues to run 100% on your computer, your data is safe in your local folder                                                | Your life's work held hostage in the cloud in proprietary data formats[^1] |
| Driving Philosophy                                         | [Privacy as a human right](https://en.wikipedia.org/wiki/Right_to_privacy) and the [Right to Repair](https://www.repair.org/) | Monetize your users                                                        |