Creativity Commons

OllamaAIModels

Learn how to install, verify, and manage Ollama and an AI model on a Mac mini, test the local API, troubleshoot common issues, and build a private, fully offline AI environment.

unsplash.com/@ditadi

Abstract

Learn how to install, verify, and manage Ollama and an AI model on a Mac mini, test the local API, troubleshoot common issues, and build a private, fully offline AI environment.

Building a Private AI Environment That Works Completely Offline

One of the biggest advantages of running AI locally is privacy. Your documents never have to leave your computer; there are no monthly API costs, and you can continue working even without an Internet connection.

This guide walks through installing:

Ollama (runs AI models locally)
Llama 3.1:8B (the AI language model, or pick another one)

By the end of this tutorial, you'll be able to verify your AI interface and do a few other nice things entirely on your Mac mini.

---

Requirements

This guide assumes:

Apple Mac M4 mini
At least 24 GB RAM

---

Step 1 – Install Ollama

Download Ollama:

https://ollama.com/download

Install it.

Once installed, open Terminal and verify it works:

%ollama --version

You should see something similar to:

%ollama version 0.xx.x

---

Step 2 – Download Llama 3.1:8B or your choice of many

In Terminal, run:

%ollama pull llama3.1:8b

This downloads roughly 5 GB.

After it finishes, test it:

%ollama run llama3.1:8b

Example:

Why is the sky blue?

The model should answer.

Exit by pressing:

Ctrl + D

or type

/bye

---

Step 4 – Start the Ollama Server

Normally, the Ollama server starts automatically.

If not, run:

%ollama serve

Leave this Terminal window open.

The API is now available at:

http://localhost:11434

---

Step 5 – Verify the API

Open a browser and visit:

http://localhost:11434

You should see a response similar to:

Ollama is running

{"models":[...]}

depending on the version.

Test Connection

Verify inside a terminal:

%ollama list

shows:

llama3.1:8b

Also verify:

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

returns your installed models.

---

Other Useful Ollama Commands

Inside a terminal: List installed models:

%ollama list

Run a model:

%ollama run llama3.1:8b

Delete a model:

%ollama rm llama3.1:8b

Update Ollama:

Download the newest installer from:

https://ollama.com/download

Show running models:

%ollama ps

Stop a running model:

%ollama stop llama3.1:8b

---

Troubleshooting

"Connection Refused"

Run:

%ollama serve

Or restart Ollama.

---

"Model Not Found"

Run:

%ollama list

Ensure the model name matches exactly.

---

Slow Responses

Try:

Closing other applications.
Using a smaller model.
Upgrading memory if possible.

---

Final Thoughts

This combination creates a capable AI environment:

Ollama serves AI models locally through a simple API.
Llama 3.1:8B offers strong performance for drafting, editing, summarizing, and brainstorming.

Because everything runs on your Mac mini, everything stays under your control; there are no per-request API charges, and you can work entirely offline once the software and model are installed.

---

Now, create a workflow and AI!

Future blogs:

1) will be about picking, installing, and working with different AI models — WooHoo!
2) will be about installing and running Localwriter (AI interface) with LibreOffice — more WooHoo!

Other AI blogs posted: