If you're running large LLMs locally, your system drive fills up fast.
In this guide, I’ll show you how to:
- Install Ollama on Windows 11
- Store ALL large model files on an external drive (Z:)
- Use a safer symbolic link method (recommended)
- Keep everything clean and future-proof
This works perfectly for large drives like a 2TB external SSD or HDD.
Why Use a Symbolic Link Instead of Environment Variables?
Ollama normally stores models in:
C:\Users\YOUR_USERNAME\.ollama
Models can easily take up:
- 4–8GB per 7B model
- 40GB+ for large models
- Hundreds of GB if you're experimenting
Instead of changing internal paths, we redirect the entire .ollama folder to:
Z:\llmmodels
Windows will silently forward all traffic there.
Ollama doesn’t even know.
Step-by-Step Installation Guide (Safe Method)
Step 1 — Plug in Your External Drive
Make sure your drive:
- Is connected
- Shows as drive letter Z or whatever letter you are using in this example, we are using Z.
- It's formatted as NTFS (recommended for Windows 11).
If Windows changes the drive letter randomly:
- Press Start
- Search "Disk Management"
- Right-click the drive
- Choose "Change Drive Letter and Paths"
- Set it permanently to Z:
Step 2 — Install Ollama Normally
Open PowerShell and run:
irm https://ollama.com/install.ps1 | iex
Once installed, DO NOT download any models yet.
Step 3 — Close Ollama Completely
Very important.
In the system tray:
- Right-click Ollama if it exist
- Click Quit
Or run:
taskkill /IM ollama.exe /F
Step 4 — Create Your Model Storage Folder on Z:
New-Item -ItemType Directory -Force "Z:\llmmodels"
You could create the folder from file explorer.
Step 5 — Delete the Default .ollama Folder (If It Exists)
Go to:
C:\Users\YOUR_USERNAME\
Delete the folder:
.ollama
If it doesn’t exist yet, that’s fine.
Step 6 — Create the Symbolic Link (The Key Step)
Open PowerShell as Administrator and run:
cmd /c mklink /J "%USERPROFILE%\.ollama" "Z:\llmmodels"
You should see:
Junction created for ...
What this does:
It tricks Windows into thinking:
C:\Users\YOU\.ollama
actually exists,
but everything is stored on:
Z:\llmmodels
Step 7 — Start Using Ollama
Now test it:
ollama pull llama3
Watch your Z: drive.
You should see storage increasing inside:
Z:\llmmodels
How to Verify It’s Working
Run:
ollama list
Then check:
Z:\llmmodels
If files are growing there — you're done.
How to Remove Models Later
List models:
ollama list
Remove one:
ollama rm llama3
Storage will be freed on Z:.