Skip to content
The GAIA installer: Node.js, the GAIA CLI and Docker installed, then GAIA downloading with a progress bar and a row per service

Your own GAIA, on your own machine

One command sets it up. Your data stays in Docker volumes on your machine, and your AI keys stay yours.

Self-hosting

Three steps to a running GAIA

From a fresh machine to your assistant in the browser.

  1. 1

    Check what you need

    • Docker. If it's missing, gaia up offers to install it.
    • 8 GB RAM, 2+ CPU cores, 20 GB free disk. The images alone take about 13 GB.
    • One AI provider. A key from OpenAI, OpenRouter or Google Gemini, any OpenAI-compatible gateway, or Ollama running locally with no key.
  2. 2

    Run one command

    Installs the GAIA CLI, then offers to run gaia up: it installs Docker if it's missing, downloads GAIA and starts it.

  3. 3

    Open GAIA and finish setup

    GAIA downloads into ~/gaia and runs in Docker. Open http://localhost:3000/setup, create the admin account and pick your AI provider. Invite others by link from the Members page in Settings.

    Keep it running with gaia update, gaia backup and gaia doctor. Every command is in the CLI reference.

Questions, answered

For personal and non-profit use, yes. GAIA charges nothing of its own; you pay your AI provider for model usage. GAIA is licensed PolyForm Noncommercial, so running it inside a business needs an enterprise licence.