Docs / Install on Linux

Install ModelBrain on Linux

Download the tarball for your CPU, unpack it, run the install script. It is a command line package: no tray icon, no windows, no root. It works with Claude Code and Cursor directly, and with ChatGPT through the relay client. x86_64 and aarch64, glibc 2.28 or newer.

Steps
  1. Download the tarball for your CPU. Get modelbrain-0.2.5-linux-x86_64.tar.gz (ordinary Intel/AMD machines) or modelbrain-0.2.5-linux-aarch64.tar.gz (ARM servers, Raspberry Pi 4/5 on a 64-bit OS) from the download page, which also lists each file's SHA-256. Check it with sha256sum -c against the .sha256 file beside it.
  2. Unpack and install. Run tar xzf modelbrain-0.2.5-linux-x86_64.tar.gz, then cd modelbrain-0.2.5-linux-x86_64 and ./install.sh --setup. That copies ModelBrain to ~/.local/lib/modelbrain, links modelbrain and connector_sync into ~/.local/bin, and registers it with Claude Code and Cursor. Leave off --setup to install without touching any client config.
  3. Check it. Run modelbrain status. It shows whether the daemon is running and which clients are set up. The daemon starts on demand the first time an assistant connects. Add --systemd to the install command to keep it running as a user service instead.
  4. Connect your assistant. Run modelbrain setup claude-code or modelbrain setup cursor if you skipped --setup, then follow Claude Code or Cursor and the quickstart.
Everyday commands
modelbrain status                     is it running, which clients are set up
modelbrain setup all                  register with Claude Code and Cursor
modelbrain daemon restart             restart the background daemon
modelbrain reasoning set-tier high    tiers: auto, minimal, default, high, off
connector_sync sync notes ~/notes     keep a folder of notes in your memory
connector_sync list                   show registered folders
ChatGPT

ChatGPT reaches ModelBrain through a hosted relay. From a terminal, run MODELBRAIN_RELAY_URL=wss://relay.modelbrain.net/device ~/.local/lib/modelbrain/modelbrain_relay_client. It prints a short pairing code. In ChatGPT, add relay.modelbrain.net/mcp as a custom connector and enter the code. Leave the command running while you use ChatGPT. Closing it disconnects ChatGPT only, not your other assistants.

Where things live
Your vault

~/.modelbrain/data, ordinary files on your disk. Set MODELBRAIN_DIR to move it. Logs are in ~/.modelbrain/logs and reasoning models, downloaded only when you turn that mode on, in ~/.modelbrain/models.

The program

~/.local/lib/modelbrain, about 200 MB. Upgrade by unpacking the new tarball and running ./install.sh again; your data is untouched. Uninstall with ~/.local/lib/modelbrain/uninstall.sh, and add --purge-data to delete the vault too.

Questions

Which Linux distributions are supported?

Any 64-bit glibc distribution with glibc 2.28 or newer, on x86_64 or aarch64: Debian 10+, Ubuntu 20.04+, RHEL, Rocky and Alma 8+, Fedora, Arch and similar. Alpine and other musl distributions are not supported. No other libraries are needed.

Is there a tray icon or window on Linux?

No. The Linux package is command line only. It works with Claude Code and Cursor directly and with ChatGPT through the relay client. There is no official Claude Desktop for Linux.

Does it need root to install?

No. It installs per user into ~/.local/lib/modelbrain and refuses to run as root.

Next

Connect ModelBrain to the assistant you actually use.

Claude Code → · Cursor →
Having trouble?

Common install and connection issues, per platform.

Troubleshooting →