Jan - Self-Hosted AI Platform

Getting Started - Docs - Changelog - Bug reports - Discord

Jan is a self-hosted AI Platform. We help you run AI on your own hardware, giving you full control and protecting your enterprises' data and IP.

Jan is free, source-available, and fair-code licensed.

Demo

👋 https://cloud.jan.ai

Features

Multiple AI Engines

• Self-hosted Llama2 and LLMs
• Self-hosted StableDiffusion and Controlnet
• Connect to ChatGPT, Claude via API Key (coming soon)
• 1-click installs for Models (coming soon)

Cross-Platform

• Web App
• Jan Mobile support for custom Jan server (in progress)
• Cloud deployments (coming soon)

Organization Tools

• Multi-user support
• Audit and Usage logs (coming soon)
• Compliance and Audit (coming soon)
• PII and Sensitive Data policy engine for 3rd-party AIs (coming soon)

Hardware Support

• Nvidia GPUs
• Apple Silicon (in progress)
• CPU support via llama.cpp (in progress)

Usage

So far, this setup is tested and supported for Docker on Linux, Mac, and Windows Subsystem for Linux (WSL).

Dependencies

• Install Docker: Install Docker here.

• Install Docker Compose: Install Docker Compose here.

• Clone the Repository: Clone this repository and pull in the latest git submodules.

  git clone https://github.com/janhq/jan.git
  
  cd jan
  
  # Pull latest submodules
  git submodule update --init --recursive

• Export Environment Variables

export DOCKER_DEFAULT_PLATFORM=linux/$(uname -m)

• Set a .env: You will need to set up several environment variables for services such as Keycloak and Postgres. You can place them in .env files in the respective folders as shown in the docker-compose.yml.

  cp sample.env .env

  Service (Docker)	env file
  Global env	.env, just run cp sample.env .env
  Keycloak	.env presented in global env and initiate realm in conf/keycloak_conf/example-realm.json
  Keycloak PostgresDB	.env presented in global env
  jan-inference	.env presented in global env
  app-backend (hasura)	conf/sample.env_app-backend refer from here
  app-backend PostgresDB	conf/sample.env_app-backend-postgres
  web-client	conf/sample.env_web-client

Install Models

wget https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/resolve/main/llama-2-7b-chat.ggmlv3.q4_1.bin -P jan-inference/llm/models

Compose Up

Jan uses an opinionated, but modular, open-source stack that comes with many services out of the box, e.g. multiple clients, autoscaling, auth and more.

You can opt out of such services or swap in your own integrations via Configurations.

• Run the following command to start all the services defined in the docker-compose.yml

# Docker Compose up
docker compose up

# Docker Compose up detached mode
docker compose up -d

• This step takes 5-15 minutes and the following services will be provisioned:

Service	URL	Credentials
Web App	http://localhost:3000	Users are signed up to keycloak, default created user is set via conf/keycloak_conf/example-realm.json on keycloak with username: username, password: password
Keycloak Admin	http://localhost:8088	Admin credentials are set via the environment variables KEYCLOAK_ADMIN and KEYCLOAK_ADMIN_PASSWORD
Hasura App Backend	http://localhost:8080	Admin credentials are set via the environment variables HASURA_GRAPHQL_ADMIN_SECRET in file conf/sample.env_app-backend
LLM Service	http://localhost:8000

Usage

• Launch the web application via http://localhost:3000.
• Login with default user (username: username, password: password)
• For configuring login theme, check out here

Configurations

TODO

Developers

Architecture

TODO

Dependencies

• Keycloak Community (Apache-2.0)
• Hasura Community Edition (Apache-2.0)

Repo Structure

Jan is a monorepo that pulls in the following submodules

├── docker-compose.yml
├── mobile-client       # Mobile app
├── web-client          # Web app
├── app-backend         # Web & mobile app backend
├── inference-backend   # Inference server
├── docs                # Developer Docs
├── adrs                # Architecture Decision Records

Common Issues and Troubleshooting

Contributing

Contributions are welcome! Please read the CONTRIBUTING.md file for guidelines on how to contribute to this project.

License

This project is licensed under the Fair Code License. See LICENSE.md for more details.

Authors and Acknowledgments

Created by Jan. Thanks to all contributors who have helped to improve this project.

Contact

For support: please file a Github ticket For questions: join our Discord here For long form inquiries: please email [email protected]

Current Features

• Llama 7Bn
• Web app and APIs (OpenAI compatible REST & GRPC)
• Supports Apple Silicon/CPU & GPU architectures
• Load balancing via Traefik
• Login and authz via Keycloak
• Data storage via Postgres, MinIO