Ollama remove model on Mac, step by step.
A single local model can take more space than every app on your Mac. Here's how to see what you have, remove what you don't use, find where the files live, and uninstall Ollama completely.
The short version
Run ollama list to see your models and their sizes, then ollama rm with a model's name, like ollama rm llama3.2. Ollama deletes the model's files unless another model shares them. On a Mac, models live in ~/.ollama/models. To uninstall Ollama, quit it, move the app to the Trash, and delete ~/.ollama.
How to remove an Ollama model
ollama list # every model, with its size
ollama ps # models loaded in memory right now
ollama stop llama3.2 # unload one from memory
ollama rm llama3.2 # delete one from disk
ollama rm llama3.2 qwen3:8b # delete severalUse the name exactly as ollama list shows it, tag included if it isn't latest: ollama rm qwen3:8b and ollama rm qwen3 are different models. Ollama needs to be running for these commands; open the app or run ollama serve.
Removing a model is safe. ollama pull downloads it again, and nothing else on your Mac depends on it. Custom models you made with ollama create are the exception: keep the Modelfile, or you'll have to recreate them by hand.
How to delete all Ollama models
There's no built-in command for it. This removes every model ollama list shows:
ollama list | awk 'NR>1 {print $1}' | xargs ollama rmRun ollama list first and make sure that's what you want.
Where are Ollama models stored on Mac?
In ~/.ollama/models, a hidden folder in your home folder. To open it in Finder, choose Go › Go to Folder and type the path.
| What | Default location | On one Mac |
|---|---|---|
| Models | ~/.ollama/models | Not installed on the Mac we measured |
| Model files | ~/.ollama/models/blobs | One sha256 file per layer |
| Model names and tags | ~/.ollama/models/manifests | Tiny |
Ollama isn't installed on the Mac this page was written on, so there are no measured sizes here. Check yours with du -sh ~/.ollama/models.
Inside, blobs holds the actual weights as files named by their hash, and manifests maps names like llama3.2:latest to those blobs. Models that share a base share blobs, which is why ollama rm sometimes frees less than the size ollama list showed. Don't delete individual blobs by hand; you'll break whichever model uses them. Use ollama rm.
To keep models somewhere else, like an external SSD, set OLLAMA_MODELS. On a Mac, Ollama's docs say to do it with launchctl, then restart Ollama:
launchctl setenv OLLAMA_MODELS "/Volumes/External/ollama-models"Existing models don't move by themselves. Copy the contents of ~/.ollama/models to the new folder first, then check ollama list.
How big Ollama models are
Roughly the parameter count times the bits per weight. At the 4-bit quantization most Ollama tags use, a 3B model is about 2 GB, an 8B model about 5 GB, a 30B model close to 20 GB, and a 70B model over 40 GB. Pull a few sizes of the same family to compare, and it's easy to pass 100 GB without noticing. ollama list shows each model's size before you decide.
How to uninstall Ollama on Mac
- Quit Ollama. Click the llama in the menu bar and choose Quit Ollama.
- Delete the app. Move
Ollama.appfrom Applications to the Trash. - Delete your models. Move
~/.ollamato the Trash. This is where the gigabytes are. - Remove what's left. Ollama's own uninstall instructions list these too:
The first line removes thesudo rm /usr/local/bin/ollama rm -rf ~/Library/Application\ Support/Ollama rm -rf ~/Library/Saved\ Application\ State/com.electron.ollama.savedState rm -rf ~/Library/Caches/com.electron.ollama ~/Library/Caches/ollama rm -rf ~/Library/WebKit/com.electron.ollamaollamacommand Ollama linked into your path. Check what each path holds before running armline.
Installed with Homebrew instead? Run brew uninstall ollama, then delete ~/.ollama. Homebrew doesn't remove your models.
Using other local AI tools too? Delete local AI models covers LM Studio, Hugging Face, and ComfyUI. Apple's own models are different: see Apple Intelligence storage.
Disk Forecast finds this for you
Disk Forecast lists your Ollama models under Cleanup › Worth a look › Ollama models, with the size of ~/.ollama/models. Nothing in Worth a look is checked until you choose it, and checking it moves every model to the Trash at once. To remove one model, use ollama rm, as the row itself explains. LM Studio and Hugging Face models get their own rows, one entry per model. It's free.
Questions
How do I remove a model from Ollama?
Run ollama rm followed by the model's name as ollama list shows it, like ollama rm llama3.2 or ollama rm qwen3:8b. Ollama deletes its files from disk unless another model shares them. ollama pull downloads it again.
How do I delete all Ollama models?
There's no single command. Run ollama list | awk 'NR>1 {print $1}' | xargs ollama rm to remove every listed model, or quit Ollama and move ~/.ollama/models to the Trash.
Where are Ollama models stored on a Mac?
In ~/.ollama/models: blobs holds the model files and manifests maps model names to them. Set OLLAMA_MODELS with launchctl setenv to store them somewhere else, then restart Ollama.
How do I uninstall Ollama on a Mac?
Quit Ollama from the menu bar, move Ollama.app to the Trash, and delete ~/.ollama, which holds the models. Ollama's docs also list /usr/local/bin/ollama and a few folders in ~/Library to remove.
Why didn't ollama rm free as much space as the model's size?
Models can share files. If two models use the same base weights, removing one keeps the shared blobs for the other. The space comes back when you remove the last model that uses them.
Related guides
Know before it's full.
Free. No account, no subscription.
macOS 14 or later · Apple Silicon and Intel · brew install --cask coreyhaines31/tap/diskforecast