Ollama can run natively on Windows, so WSL is not required. After installation, the

ollama

http://localhost:11434

1. Check if Ollama is installed

Open CMD:

ollama --version

If Ollama is installed, you should see something similar to:

ollama version ...

You can also check whether Windows can find the executable:

where ollama

Example:

C:\Users\YourName\AppData\Local\Programs\Ollama\ollama.exe

If you get:

'ollama' is not recognized as an internal or external command

Ollama is either not installed or its PATH is not available in the current terminal.

2. Check whether Ollama is already running

Run:

ollama ps

You can also test the API:

curl http://localhost:11434/api/tags

If Ollama is running, you should get JSON containing your installed models.

Ollama's Windows application normally runs in the background after installation. (Ollama)

3. Start Ollama manually from CMD

If Ollama isn't running, use:

ollama serve

You may see something like:

Listening on 127.0.0.1:11434

Important: keep this CMD window open while using the manually started server.

Ollama officially provides

ollama serve

Better workflow

Open CMD Window 1:

ollama serve

Then open CMD Window 2 and use:

ollama ps

or:

ollama ls

4. Check all installed models

Use:

ollama ls

You may see:

NAME              ID              SIZE      MODIFIED
gemma3:latest     abc123...       3.3 GB    ...
qwen3:8b          xyz456...       5.2 GB    ...

Older guides may show:

ollama list

but the current CLI documentation uses:

ollama ls

(Ollama)

5. Check currently running models

Use:

ollama ps

Example:

NAME          ID        SIZE      PROCESSOR    UNTIL
qwen3:8b      abc...    5.2 GB    100% GPU     ...

This is different from

ollama ls

CodeCopy

ollama ls

Shows models installed on your computer.

CodeCopy

ollama ps

Shows models currently loaded/running.

(Ollama)

6. Get information about a model

For example:

ollama show qwen3:8b

Or:

ollama show gemma3

This can be useful for checking the model's configuration and information.

7. Download/install a new model

Use:

ollama pull MODEL_NAME

For example:

ollama pull gemma3

Or:

ollama pull qwen3:8b

Ollama downloads the model to your local machine. (Ollama)

After downloading, verify:

ollama ls

8. Run a model

After installing a model:

ollama run gemma3

For example:

ollama run qwen3:8b

You'll enter an interactive chat:

>>> Hello

Then Ollama responds locally.

The current CLI uses

ollama run MODEL_NAME

9. Run a model with a prompt directly

You don't have to enter interactive mode.

Example:

ollama run gemma3 "Explain Docker in simple words"

Or:

ollama run qwen3:8b "Write a Python hello world program"

10. Stop a running model

First check:

ollama ps

Then:

ollama stop MODEL_NAME

Example:

ollama stop qwen3:8b

The official CLI provides

ollama stop

11. Remove/delete a model

First check the exact model name:

ollama ls

Then:

ollama rm MODEL_NAME

Example:

ollama rm qwen3:8b

Then verify:

ollama ls

The model should no longer appear.

Ollama's current CLI documents

ollama rm

12. Replace an old model with a new one

For example, suppose you currently have:

gemma3
qwen3:8b
llama3.2

First inspect:

ollama ls

Remove something you don't need:

ollama rm llama3.2

Install your new model:

ollama pull qwen3.5

Then run it:

ollama run qwen3.5

13. Update an existing model

You can simply pull it again:

ollama pull MODEL_NAME

Example:

ollama pull gemma3

Ollama can download the newer/different layers rather than unnecessarily downloading everything again. (Ollama Documentation)

14. See all available Ollama commands

Run:

ollama

This opens Ollama's interactive CLI menu in current versions. (Ollama)

Or use:

ollama --help

For a specific command:

ollama serve --help

ollama run --help

ollama pull --help

15. Check the Ollama API

Ollama normally exposes:

http://localhost:11434

Test it:

curl http://localhost:11434

You can also check installed models through the API:

curl http://localhost:11434/api/tags

If the server is working, you'll receive JSON.

16. Completely remove a model

If you want to remove a model and its downloaded data:

ollama rm MODEL_NAME

Example:

ollama rm gemma3

Then:

ollama ls

If it isn't listed anymore, the model has been removed.

17. Completely uninstall Ollama from Windows

If you want to remove Ollama itself, don't just delete

ollama.exe

Go to:

Windows Settings → Apps → Installed apps → Ollama → Uninstall

The official Windows installer registers an uninstaller in Windows. (Ollama)

Ollama's Windows files include locations such as:

%LOCALAPPDATA%\Ollama

%LOCALAPPDATA%\Programs\Ollama

and models/configuration are normally under:

%HOMEPATH%\.ollama

(Ollama)

18. Reinstall Ollama

The easiest method is the official Windows installer:

Download Ollama for Windows

After installation, open a new CMD window:

ollama --version

Then:

ollama ls

Then install a model:

ollama pull gemma3

And run it:

ollama run gemma3

19. One complete CMD workflow

If you're setting up Ollama from scratch, this is the basic sequence:

ollama --version

ollama ls

ollama ps

If the server isn't running:

ollama serve

Open another CMD:

ollama ls

Install a model:

ollama pull gemma3

Check it:

ollama ls

Run it:

ollama run gemma3

After you're finished:

ollama ps

Stop the model:

ollama stop gemma3

If you don't want the model anymore:

ollama rm gemma3

20. Quick command cheat sheet

|Purpose|Command| |---|---| |Check installation|

ollama --version

where ollama

ollama serve

ollama ls

ollama ps

ollama pull MODEL

ollama run MODEL

ollama stop MODEL

ollama rm MODEL

ollama show MODEL

ollama --help

curl http://localhost:11434

curl http://localhost:11434/api/tags

These commands correspond to the current Ollama CLI reference. (Ollama)

Recommended workflow

If your goal is simply "check what I have → remove old models → install a new model", use:

ollama --version

ollama ls

ollama ps

Then remove unwanted models:

ollama rm MODEL_NAME

Install the new one:

ollama pull NEW_MODEL

Verify:

ollama ls

Run:

ollama run NEW_MODEL

And when finished:

ollama stop NEW_MODEL

Note: On Windows, Ollama normally runs as a background application, so you usually don't need to manually run

ollama serve