← Mac guides

How to stop Ollama running in the background on Mac

Ollama keeps models warm in Apple silicon unified memory and can auto-start via its menu bar app or Homebrew’s launch agent. Here is how to stop it cleanly—and bring it back when you need it.

Updated 2026-11-19 · macOS 13+ · Works for Ollama.app and Homebrew ollama

1. Check which models are loaded in RAM right now

Before killing the server, check what Ollama currently holds in GPU/unified memory:

ollama ps

To unload one specific model immediately without stopping the daemon:

ollama stop <model-name>

2. Stop Ollama completely (menu bar app vs. Homebrew)

How you stop Ollama depends on how you installed it on macOS:

  • Installed as Ollama.app (DMG / Cask): Click the llama icon in the macOS menu bar and choose Quit Ollama. Closing a terminal window does not stop the menu bar app.
  • Installed as a Homebrew service: Stop the background LaunchAgent so macOS does not immediately respawn ollama serve:
brew services stop ollama

Started manually via ollama serve: Press Control + C in that terminal tab, or verify nothing remains running:

pgrep -fl ollama
Why Ollama comes right back: If you only run killall ollama while Homebrew’s LaunchAgent is loaded, launchd restarts Ollama within seconds. Stopping the Homebrew service—or pausing Ollama through AI Stopper—prevents that respawn loop.

3. Pause Ollama (and LM Studio, Jan, llama.cpp) in one click

AI Stopper recognizes Ollama.app, the ollama CLI binary, and the ollama Homebrew service natively—alongside LM Studio, Msty, Jan, AnythingLLM, llama.cpp, and 27 other Mac AI tools.

  • On Pause: Stops Ollama processes and runs the clean Homebrew service stop if it was active, while Restart Guard keeps it off.
  • On Resume: Restarts your Homebrew Ollama service automatically if it was running before the pause.

Want to reclaim RAM across local models, Whisper apps, and coding IDEs together? See our Save Mac RAM & Battery workflow.