- What Ollama is and why people use it
- How to install Ollama on Mac, Windows and Linux
- How to run your first model in Ollama
- The Ollama commands you'll use most
- How to customise a model with an Ollama Modelfile
- Connecting Ollama to other apps and coding tools
- Ollama tips for privacy and safety
- Is Ollama right for your business?
- Frequently asked questions
To use Ollama, install it from ollama.com or its GitHub page, open a terminal and type ollama run followed by a model name, such as gemma4. The model downloads once, then runs on your own computer. The detail people miss: bigger models need more memory, so start small.
Ollama is the tool I point people to when they ask, 'Can I run my own AI on my laptop?' The answer is yes, and Ollama makes it surprisingly simple. You type one command, a model downloads, and you're chatting with an AI that runs on your own machine.
This guide explains what Ollama is, how to install it on Mac, Windows and Linux, the handful of commands you'll use every day, and how a small business might put a local model to work. Every step comes from Ollama's own GitHub page and documentation.
What Ollama is and why people use it
Ollama is a free, open-source tool for running AI models. Its GitHub page says it helps you get up and running with models such as Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen and Gemma, and lets you access the latest releases locally or in the cloud.
The appeal is control. When a model runs locally, your prompts stay on your computer rather than going to a chatbot company. You can also try lots of open models side by side, and connect them to other apps through Ollama's built-in API.
- Privacy: useful when you're working with drafts or notes you'd rather not upload.
- Choice: swap between open models from different makers in seconds.
- Offline work: once a model is downloaded, it runs on your machine.
- Building: developers can plug Ollama into their own apps and coding tools.
How to install Ollama on Mac, Windows and Linux
Ollama's GitHub page gives two ways to install on each system: a one-line command or a normal download.
- Mac: open Terminal and paste the one-line curl install command from Ollama's download page, or download the Ollama.dmg file and install it like any other app.
- Windows: open PowerShell and paste the one-line install command from Ollama's download page, or download OllamaSetup.exe and run it.
- Linux: paste the same curl install command into a terminal, or follow the manual install guide.
- Docker: if you use containers, there's an official ollama/ollama image on Docker Hub.
A quick word of caution I'd give anyone: only run install commands copied from the official site or GitHub page, never from a random forum post or video description.
How to run your first model in Ollama
Once it's installed, you talk to Ollama from the terminal. Typing ollama on its own offers to run a model or connect Ollama to apps. To chat straight away, the docs use this example:
- Open your terminal or command prompt.
- Type ollama run gemma4 and press Enter.
- Wait while the model downloads the first time.
- Type your question when the prompt appears, and press Enter to get an answer.
- Browse the model library at ollama.com/library to find other models to try.
Bigger models give better answers but need a more powerful computer. If a model feels slow, try a smaller version from the library.
The Ollama commands you'll use most
You only need a few commands. Here's the list, as Ollama's CLI reference describes each one:
| Ollama command | What it does |
|---|---|
| ollama run gemma4 | Starts a chat with a model, downloading it first if needed |
| ollama pull gemma4 | Downloads a model without starting a chat |
| ollama ls | Lists the models you've downloaded |
| ollama ps | Shows which models are running right now |
| ollama stop gemma4 | Stops a running model |
| ollama rm gemma4 | Deletes a model to free up space |
| ollama create -f Modelfile | Builds your own customised model |
How to customise a model with an Ollama Modelfile
A Modelfile is a small text file that tells Ollama how to set up your own version of a model. The docs show an example that starts with a line saying which model to build from, followed by a SYSTEM line giving it standing instructions.
For a small business, that's a neat trick. You could create a model that always writes in British English, keeps answers short, and follows your house style, then run it with one command. Run ollama create with your file, and your custom model appears in your list.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Connecting Ollama to other apps and coding tools
Ollama runs a local server on your machine, at port 11434 on localhost, which other apps can talk to. It offers a REST API, including a chat endpoint, plus libraries for Python and JavaScript. That's how chat apps and automation tools can use your local models.
The newer ollama launch command sets up supported apps such as OpenCode, Claude Code, Codex and VS Code to use Ollama models. If you're exploring AI for development work, my guide to speeding up coding work with AI prompts has ideas that work with local models too.
Ollama tips for privacy and safety
- Download models from Ollama's library or other sources you trust.
- Check each model's licence before using it for commercial work.
- Remember that any app you connect to Ollama may have its own data rules.
- Review AI-written code before running it, and never paste passwords or API keys into prompts.
- Keep your computer and Ollama updated.
Running locally is one way to keep things private, but it isn't the only one. My guide on whether ChatGPT is private explains the settings to check if you use a hosted chatbot too.
Is Ollama right for your business?
Ollama suits people who are comfortable with a terminal, care about privacy, or want to experiment with open models. If you'd rather click than type, a desktop app might suit you better. And if you're not sure whether local AI is worth it for your business at all, that's exactly the kind of question I help with as an AI consultant.
Frequently asked questions
Is Ollama free?
Yes. Ollama is free, open-source software you can download for Mac, Windows and Linux. Models in its library each come with their own licence.
Does Ollama work offline?
Once you've downloaded a model, Ollama runs it on your own computer. You'll need a connection to download models and updates.
What models can I run with Ollama?
Ollama's GitHub page mentions models such as Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen and Gemma. Browse ollama.com/library for the full range.
How do I delete a model in Ollama?
Run ollama rm followed by the model's name. Use ollama ls first to see the models you have.
Can other apps use my Ollama models?
Yes. Ollama runs a local API at localhost that apps can connect to, and the ollama launch command can set up supported coding tools for you.
