Before pulling a multi-gigabyte model, prove Ollama is installed and responding.
Windows
- Press Win, type PowerShell, open Windows PowerShell (or Terminal).
- Run:
ollama --version
- Optional second check:
ollama list
An empty list is fine — you haven't pulled a model yet.
You should see: a version number (e.g. ollama version is 0.x.x) with no "command not found" error.
If ollama isn't recognised: close and reopen PowerShell after install, or reboot once. Confirm the tray icon is present.
macOS
- Open Terminal (Spotlight → type Terminal).
- Run:
ollama --version
- Optional:
ollama list
You should see: a version string. If you use the menu-bar app, the icon should be visible without running serve manually.
Linux
- Open your terminal.
- Run:
ollama --version
- If that fails with "connection refused" when you pull later, start the service:
sudo systemctl start ollama
(or run ollama serve in a separate terminal for this session)
You should see: version output; ollama list returns an empty table or headers — not a crash.
If you hit "command not found" on Windows, nine times out of ten it's just that the installer finished while your terminal was already open. Close the window, open a fresh one, and try again before you panic.
Continue — download your starter model.