Install LocalAI
Start with the install script on Linux, or on a Mac with Apple Silicon. On Windows, on an Intel Mac, in a cluster, or when you want to keep LocalAI apart from the host, use a container. On a Mac you can also use the app.
| You have | Use | Command or guide |
|---|---|---|
| Linux, or a Mac with Apple Silicon | Install script | curl -sSL https://localai.io/install.sh | sh |
| Windows (WSL2), an Intel Mac, a server, or a GPU image | Container | Containers |
| A Mac and you want an app | macOS app | macOS |
| Linux with a desktop | Launcher or binary | Linux |
| A cluster | Helm chart | Kubernetes |
| A change to the code | Source | Build from source |
Install script
The script downloads the release binary for your machine, checks it against the checksums of the release, and puts it on your PATH. It builds nothing. Running it twice changes nothing.
The API and the web interface are then at http://localhost:8080.
You can set these environment variables before the command:
| Variable | Effect |
|---|---|
LOCALAI_VERSION | Install a given release, for example v4.7.1. The default is the latest release. |
INSTALL_DIR | Install into this folder. The default is /usr/local/bin, or ~/.local/bin when /usr/local/bin is not writable. |
LOCALAI_FORCE | Set to 1 to install again when the same build is present. |
LOCALAI_NO_SUDO | Set to 1 to never ask for sudo. LocalAI then installs into ~/.local/bin. |
The script supports Linux and macOS on amd64 and arm64. There is no binary for Intel Macs. Use a container there. For a different install, see the other guides.
Container
Podman works the same way. For GPU images, see Containers.
macOS app
Download LocalAI.dmg, open it and move LocalAI to Applications. The DMG is signed and notarized. See macOS for the details.
What next
Run a first model with the quickstart.