How to pause Ollama on Mac
Ollama keeps multi-gigabyte models loaded in Apple silicon unified memory and can silently restart via Homebrew’s LaunchAgent. Here is how to stop it reliably and restore it automatically.
What AI Stopper detects for Ollama
- macOS app bundle:
Ollama.app(menu bar server) - CLI command:
ollama - Homebrew service:
ollama(stopped cleanly during pause and restarted on resume if it was running)
How to stop Ollama manually on macOS
- Run
ollama psto see which models currently occupy unified memory, and runollama stop <model>to unload one. - If you installed the DMG/Cask app, click the Ollama llama icon in the menu bar and select Quit Ollama.
- If you run Ollama via Homebrew services, run
brew services stop ollamasolaunchddoes not immediately respawn it.
Check for running processes in Terminal (read-only):
ollama ps 2>/dev/null; pgrep -fl ollama
Stop Homebrew service:
brew services stop ollama
Why manual quitting misses edge cases: Running killall ollama without stopping the Homebrew service only pauses Ollama for a few seconds before macOS relaunches it. Stopping the service cleanly is essential to keep RAM and battery free.
What happens when you pause Ollama with AI Stopper
- Stops
Ollama.app, directollama serveprocesses, and theollamaHomebrew service together. - Remembers whether your Homebrew Ollama service was active and restarts it automatically when you click Resume Tools.
- Pairs cleanly with LM Studio, Jan, Msty, and llama.cpp when you want all local LLMs off while unplugged.