Ollama + Gemma 4 Local AI Setup Guide

Ollama + Gemma 4

Local AI Setup Guide

Install and run a local LLM on an external drive — step by step

 

Overview

This guide walks you through installing Ollama and the Gemma 4 language model entirely on an external drive, so your main system drive stays clean. You will also learn how to expose the model via an API for use in tools like Postman or your own applications.

 

Topic

Details

Tool

Ollama — a lightweight local LLM runner

Model

Gemma 4 by Google (~9 GB RAM required)

Storage

External drive (e.g. D:\ or F:\

OS

Windows (PowerShell)

API Port

localhost:11434

 

Part 1 — Download Ollama

Visit the official Ollama website to download the Windows installer.

 

•        Go to: https://ollama.com/download

•        Download the OllamaSetup.exe file

 

📝  Note: The installer file is approximately 1.7 GB. Make sure you have a stable internet connection before downloading.

 

Part 2 — Install Ollama to Your External Drive

Instead of double-clicking the installer (which installs to C:\ by default), use PowerShell with the /DIR flag to choose your external drive location.

 

2.1  Open PowerShell and navigate to the installer

Open PowerShell and change directory to wherever you saved OllamaSetup.exe. Then run the command below, replacing D:\Ollama with your own drive letter and folder:

 

.\OllamaSetup.exe /DIR="D:\Ollama"

 

📝  Note: Replace D:\Ollama with your external drive path, e.g. F:\Ollama if your drive is on F:\

 

 

After installation, you should be able to see the Ollama folder extracted on your external drive (D:\ in this example):

 

 

Complete the installation wizard to finish setup.

 

Part 3 — Store Models on Your External Drive

By default Ollama saves downloaded models to your C:\ drive. You need to point it to your external drive so models are stored there instead.

 

3.1  Set the Model Storage Path in Settings

Open Ollama settings and update the "Model location" to your external drive path:

 

 

Part 4 — Add Ollama to Environment Variables

Adding Ollama to your Windows environment variables lets you run ollama commands from any folder in PowerShell.

 

1.     Press Win + R, type sysdm.cpl, and press Enter

2.     Click the Advanced tab, then Environment Variables

3.     Under User variables, click New

4.     Set Variable name to: OLLAMA_MODELS

5.     Set Variable value to: D:\Ollama (or your external drive path)

6.     Click OK, then restart Ollama

 

 

Part 5 — Verify the Installation

After setting the environment variable, fully close and reopen PowerShell — existing sessions do not pick up PATH changes.

 

📝  Note: Always close and reopen PowerShell after changing environment variables. Otherwise the new PATH will not be active.

 

Run the following command to confirm Ollama is installed correctly:

 

ollama --version

 

You should see the Ollama version number printed in the terminal:

 

 

Part 6 — Install Gemma 4 Model

Now that Ollama is set up, you can pull and run the Gemma 4 model. Gemma 4 is a powerful open model made by Google.

 

6.1  Find the install command

Visit the Ollama library page for Gemma 4 to get the exact pull command:

 

•        URL: https://ollama.com/library/gemma4

 

 

6.2  Run the pull command in PowerShell

Copy the command shown on the page and run it in PowerShell:

 

 

📝  Note: Gemma 4 requires approximately 9 GB of available RAM to run. Close other heavy applications before starting the model.

 

Once the download completes, you will see the model files stored in your external drive folder:

 

 

Part 7 — Chat with Gemma 4

You can interact with Gemma 4 directly in the terminal using the ollama run command:

 

ollama run gemma4

 

This opens an interactive chat session in PowerShell:

 

 

Part 8 — Access Gemma 4 via API

Ollama exposes a local REST API on port 11434 by default. You can use this to integrate Gemma 4 with your own code, Postman, or any HTTP client.

 

8.1  Expose the port (VS Code example)

In Visual Studio Code, open the Ports panel and add port 11434. Set visibility to Public to get a remote URL you can share or use externally:

 

 

8.2  Send a request from Postman

Use the following endpoint with a JSON payload to query Gemma 4 from Postman or any API client:

 

POST http://localhost:11434/api/generate

 

Example payload and response in Postman:

 

 

Quick Reference — Command Cheat Sheet

Action

Command

Install to external drive

.\OllamaSetup.exe /DIR="D:\Ollama"

Check Ollama version

ollama --version

Pull Gemma 4 model

ollama pull gemma4

Run Gemma 4 in terminal

ollama run gemma4

List installed models

ollama list

API endpoint

http://localhost:11434/api/generate

 

Comments

Post a Comment

Popular posts from this blog

How to enable Google Sheet in Google Cloud project

Image
 Hi, In this post, we will see how to enable Google sheet APis in Google Cloud project. Go to Google Cloud Console: Visit: https://console.cloud.google.com/ Make sure you're in your project Navigate to APIs & Services: Go to "APIs & Services" > "Library" (or use the search) Enable Google Sheets API: Search for "Google Sheets API" Click on it Click "Enable" Wait a few minutes: After enabling, wait 1–2 minutes for the change to propagate. Thank you.
Read more

Finding Your Redis Credentials: Host, Port, and Password

 Finding Your Redis Credentials: Host, Port, and Password Here is how to hunt down your Redis credentials in seconds. 1. The Host: Where is Redis living? In 90% of local development scenarios (including WSL), Redis lives right on your machine. Standard Local Host: 127.0.0.1 or localhost Remote Server: If you're connecting to a staging or production server, you will need that server's specific IP address or domain name (e.g., redis.example.com ). 2. The Port: Which door is open? The default port for Redis is 6379 . However, if you're running multiple instances, it might be different. To confirm exactly which port your instance is listening on, run: redis-cli CONFIG GET port Alternatively, you can ask your system which process is "squatting" on the network: sudo ss -tlnp | grep redis If you see 0.0.0.0:6379 , you're using the default. 3. The Password: Is the gate locked? By default, a fresh Redis installation has no password . In a local environment, this is ...
Read more

Ways to install your own extension

 In visual studio  Extension Manager ,  shows extensions from the Marketplace. But you can install your own .vsix extension manually — here’s how: 🛠️ Ways to install your own extension Double‑click the VSIX file : After you build your extension project, Visual Studio produces a .vsix file. Just double‑click it, and the Visual Studio Installer will open to install it. Use the Experimental Instance : When debugging your extension, Visual Studio launches a special “Experimental Instance” where your extension is auto‑installed for testing. Command line install : You can use VSIXInstaller.exe (included with Visual Studio) to install your extension manually. Example:                          VSIXInstaller.exe MyExtension.vsix Local install via Extensions menu : In newer Visual Studio versions, go to Extensions → Manage Extensions → Install from VSIX… and select your file.
Read more