AI Setup Studio
Local AI

How to Run AI Locally With Ollama

Install Ollama, download your first model, and chat with AI privately on your computer.

9 min read Intermediate Reviewed 2026-10

Ollama lets you download and run open AI models directly on your own computer. Your prompts stay on your machine, and once a model is downloaded it can work offline.

Performance depends on your hardware, especially memory (RAM) and graphics card. This guide helps you install Ollama and pick a model that fits.

What you need

  • macOS, Windows, or Linux
  • Memory: as a rule of thumb from the Ollama project, about 8 GB RAM for 7B models, 16 GB for 13B models, and 32 GB for 33B models
  • Free disk space: models typically need several gigabytes each
  • Optional: a supported GPU for faster responses

Step-by-step

  1. 1

    Download Ollama (macOS and Windows)

    Go to ollama.com/download, choose your operating system, and run the installer. Ollama runs in the background after installation.

  2. 2

    Install on Linux

    On Linux, the official install script is the usual method. Review scripts before running them if you prefer.

    curl -fsSL https://ollama.com/install.sh | sh
  3. 3

    Confirm it works

    Open a terminal (Terminal on macOS, PowerShell on Windows) and check the version.

    ollama --version
  4. 4

    Run your first model

    Download and start a small, capable model. The first run downloads the model, which can take a few minutes.

    ollama run llama3.2

    Browse the model library on ollama.com to see other models and their sizes. Choose smaller models if your computer has limited memory.

  5. 5

    Chat

    Type a question at the prompt and press Enter. To exit, type /bye.

  6. 6

    Manage models

    List installed models, download another one, or remove one you no longer need to free disk space.

    ollama list
ollama pull mistral
ollama rm mistral
  7. 7

    Use the local API (optional)

    Ollama serves a local API on your computer at http://localhost:11434, which other apps and chat interfaces can connect to.

    curl http://localhost:11434/api/generate -d '{"model": "llama3.2", "prompt": "Hello!"}'

Common errors

"address already in use" when running ollama serve

Ollama is probably already running in the background (for example from the desktop app). You can use it directly without starting it again.

Responses are very slow

Try a smaller model, close memory-heavy apps, and check whether your GPU is supported. Large models on limited RAM will be slow.

Model download fails or stalls

Check your internet connection and free disk space, then run the pull command again. Downloads resume where possible.

"ollama: command not found"

Open a new terminal after installing. On macOS, make sure the Ollama app has been opened at least once.

Tips

  • Start with a small model and move up if your hardware handles it well.
  • Local models are generally less capable than the largest cloud models, but great for privacy.
  • Pair Ollama with a chat interface for a ChatGPT-like experience.

Official resources

FAQ

AI Setup Studio is an independent technical services company. We are not affiliated with, endorsed by, or sponsored by OpenAI, Anthropic, Google, Microsoft, GitHub, Cursor, Ollama, n8n, Make, or any other AI tool vendor unless expressly stated. Product names are trademarks of their respective owners and are used only to describe compatibility.

ServicesGet My AI Set Up