Running out of disk space is the number one reason people search for how to use the ollama delete model command. Each model you pull can weigh anywhere from 2 GB to 40 GB, and Ollama never removes anything on its own. If you are running local AI models on a laptop or a workstation with a modest SSD, old experiments pile up fast.
The good news is that cleanup takes one command. Whether you are on Windows PowerShell, macOS Terminal, or a Linux shell, the workflow is identical: list what you have, remove what you no longer need, and free the space immediately.
This guide covers deleting a single model, wiping every model at once, finding the raw files on disk, cleaning up orphaned blobs, and fully uninstalling Ollama per operating system.
ollama list to see installed models, then run ollama rm <model> — for example ollama rm llama3.1 — to delete it. To remove all models, pipe the list into a loop or delete the models folder manually at %USERPROFILE%.ollamamodels on Windows or ~/.ollama/models on macOS and Linux.Step 1: List Your Installed Models
Before deleting anything, see exactly what is installed and how much space each model takes. Run this in PowerShell, Terminal, or any shell:
ollama list
The output shows the model name, tag, size, and when it was last modified. Copy the exact name from the NAME column — you will need it in the next step, including the tag if one is shown (for example llama3.1:8b).
Step 2: Delete a Single Ollama Model
The core ollama delete model command is ollama rm followed by the model name. It works identically in PowerShell and bash:
ollama rm llama3.1
If the model has a specific tag, include it to target that exact version:
ollama rm llama3.1:8b
Ollama confirms the removal and reclaims the disk space right away. Run ollama list again to verify the model is gone.
If You Get “Model Not Found”
The name must match exactly what ollama list shows, and the match is case sensitive on some systems. When in doubt, copy the name straight from the list output instead of typing it from memory.
Step 3: Delete All Ollama Models at Once
Ollama has no built-in “remove all” flag, so you loop over the list. The loop differs slightly between PowerShell and bash.
PowerShell (Windows)
ollama list | Select-Object -Skip 1 | ForEach-Object { ollama rm ($_ -split 's+')[0] }
Bash (macOS and Linux)
ollama list | tail -n +2 | awk '{print $1}' | xargs -I {} ollama rm {}
Both commands skip the header row, extract each model name, and pass it to ollama rm. Check your disk space afterwards — large models free up several gigabytes each.
If you are cleaning up because of VRAM pressure rather than disk space, our guide to Ollama GPU and memory errors explains how to pick smaller quantizations instead of deleting models outright.
Where Ollama Stores Model Files
Ollama keeps everything under a .ollama folder in your user profile. Knowing the location helps when the CLI cannot start or when you want to reclaim space manually.
| Operating system | Model storage path |
|---|---|
| Windows | %USERPROFILE%.ollamamodels |
| macOS | ~/.ollama/models |
| Linux | ~/.ollama/models (or /usr/share/ollama/.ollama/models for the systemd service) |
Inside you will find two subfolders: manifests holds the model metadata, and blobs holds the actual weight files. The blobs are what consume gigabytes of space.
Manual Blob Cleanup
Occasionally ollama rm leaves orphaned blobs behind, especially after interrupted pulls or manual file edits. The safest fix is to delete the entire models folder after removing all models through the CLI.
PowerShell (Windows)
Remove-Item -Recurse -Force "$env:USERPROFILE.ollamamodels"
Bash (macOS and Linux)
rm -rf ~/.ollama/models
Do not delete individual blob files while Ollama still has models installed — several models can share the same blob, so removing one file can silently corrupt other models. Always use ollama rm first and only wipe the folder as a final cleanup step.
How to Fully Uninstall Ollama
If you want Ollama gone entirely, removing the models is only half the job. Each operating system has its own uninstall steps, and leaving the .ollama folder behind keeps gigabytes of data on your drive.
Windows
Open Settings, go to Apps, find Ollama, and click Uninstall. Then remove the leftover data folder in PowerShell:
Remove-Item -Recurse -Force "$env:USERPROFILE.ollama"
Also delete %LOCALAPPDATA%ProgramsOllama if the uninstaller left it behind. Restart to clear any background process.
macOS
If you installed the app, drag Ollama from Applications to the Trash. If you installed via Homebrew, run:
brew uninstall ollama
rm -rf ~/.ollama
Linux
Stop and disable the systemd service first, then remove the binary, service file, user, and data:
sudo systemctl stop ollama
sudo systemctl disable ollama
sudo rm /etc/systemd/system/ollama.service
sudo rm $(which ollama)
sudo rm -rf /usr/share/ollama
sudo userdel ollama
Finally remove any user-level data with rm -rf ~/.ollama. Run sudo systemctl daemon-reload so systemd forgets the deleted service.
Conclusion
Deleting Ollama models is a two-command job: ollama list to see what you have, ollama rm <model> to remove it. For bulk cleanup, loop over the list in PowerShell or bash, or delete the models folder directly after the CLI removals.
If you are reinstalling after a cleanup, our Ollama API server setup guide walks you through a fresh, properly configured installation. Keep only the models you actually use, and your local AI setup will stay fast and lean.