Run local AI on Linux

1. Check memory and architecture

Open Terminal and run:

free -h
uname -m

x86_64 is the common Intel/AMD desktop architecture; aarch64 is ARM64. Match your app download to it. Use the memory guide to choose a small model.

2. Install the Hugging Bay app

The desktop page checked on 17 October 2026 offers x86_64 AppImage and .deb packages for Ubuntu 22.04+ or Debian 12+. Choose the AppImage if you want to avoid an admin install.

Download it, follow the hash and signature checks beside that exact file, then use the page's command to make it executable and launch it. Follow the first-chat guide. The .deb alternative uses the install command on the download page. Do not use an x86_64 package on an ARM machine.

3. Or install Ollama

The official Linux installer is:

curl -fsSL https://ollama.com/install.sh | sh

This runs Ollama's installer on your computer and may request administrator access. Review its Linux instructions first. Then run ollama --version and continue with the Ollama guide.

If the service is not running, use ollama serve in a separate terminal. If it says the address is already in use, do not start a second server.

Other routes

For a graphical runner, download the appropriate LM Studio AppImage, make that downloaded file executable in its Properties dialog, and open it. Check the current LM Studio requirements, then follow the chat guide.

For a native terminal program, follow hbay.