How to Install Ollama on Mac — Step by Step Guide (2026)

Installing Ollama on a Mac is the single best thing I’ve done for my AI workflow in the past two years. The reason is simple: Apple Silicon Macs are genuinely excellent at running local AI models. The M1, M2, and M3 chips have unified memory — shared between the CPU and GPU — which means a MacBook with 16 GB of RAM can access all 16 GB for AI inference, unlike a Windows PC where your GPU might have only 6 GB of dedicated VRAM.

In plain terms: your Mac is probably better at running local AI than you think.

This guide will walk you through the complete process of installing Ollama on macOS — whether you have an Apple Silicon Mac (M1/M2/M3/M4) or an older Intel Mac — and running your first AI model within 10 minutes.

Apple Silicon vs Intel Mac — What You Need to Know First

Before we install anything, let’s confirm which type of Mac you have. This matters because Ollama behaves differently and performs differently depending on your chip.

How to check your Mac’s chip

Click the (Apple) menu in the top-left corner → select About This Mac. Look at the Chip or Processor line:

• Shows Apple M1 / M2 / M3 / M4 → You have Apple Silicon — Ollama works brilliantly on your Mac
• Shows Intel Core i5 / i7 / i9 → You have an Intel Mac — Ollama still works, CPU-only with lower speed

Apple Silicon Mac — Why it’s the best local AI machine

Apple Silicon’s unified memory architecture means there’s no separate GPU VRAM limit. The GPU and CPU share the same memory pool. On a MacBook Air M2 with 16 GB RAM, Ollama can use the full 16 GB for model inference, accelerated by the GPU — giving you desktop-class AI speed in a fanless laptop.

Intel Mac — what to expect

Intel Macs don’t have the same unified memory advantage. Ollama runs entirely on the CPU, which means slower responses — typically 3–10 tokens per second depending on your processor and the model size. It still works, and for lighter models like Phi-3 Mini it’s perfectly usable.

macOS Requirements for Ollama

• macOS version: macOS 11 Big Sur or later (macOS 12 Monterey or newer recommended)
• RAM: 8 GB minimum (16 GB recommended for a comfortable experience)
• Storage: 10 GB+ free space (models range from 2 GB to 40 GB each)
• Chip: Apple Silicon (M1–M4) or Intel Core — both supported

Check your macOS version: Click the Apple menu → About This Mac → look at the macOS version number shown.

Step-by-Step: How to Install Ollama on Mac

Step 1 — Go to the Official Ollama Download Page

Open Safari or any browser and navigate to:

https://ollama.com/download

The macOS tab should be selected by default when visiting on a Mac. If not, click macOS.

Step 2 — Download the Mac Installer

Click the “Download for macOS” button. This downloads Ollama-darwin.zip — a compressed application package. The file is around 100–200 MB.

Step 3 — Install Ollama

1. Open your Downloads folder in Finder
2. Double-click Ollama-darwin.zip to extract it — you’ll get Ollama.app
3. Drag Ollama.app to your Applications folder
4. Open your Applications folder and double-click Ollama to launch it
5. If macOS shows a security warning: “Ollama cannot be opened because it is from an unidentified developer” — go to System Settings → Privacy & Security → scroll down and click “Open Anyway” next to the Ollama mention

Once launched, Ollama runs in the background. You’ll see the small Ollama llama icon in your menu bar (top-right area of your screen).

Alternative — Install via Terminal (One Command)

If you’re comfortable with the Terminal, this is the fastest installation method:

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

Open Terminal (find it in Applications → Utilities, or press Cmd + Space and type “Terminal”), paste the command above, and press Enter. Ollama installs and starts automatically.

Step 4 — Install via Homebrew (for developers)

If you use Homebrew as your Mac package manager, Ollama is available as a cask:

brew install ollama

The Homebrew method makes updating Ollama easier later — just run brew upgrade ollama to stay on the latest version.

Step 5 — Verify the Installation

Open Terminal and run:

ollama --version

You should see output like ollama version 0.6.2. Also check your menu bar — the Ollama llama icon confirms the background service is running.

If you get “command not found”, try opening a new Terminal window. If it still doesn’t work, run this to add Ollama to your PATH:

echo 'export PATH="/usr/local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc

Download and Run Your First AI Model on Mac

Now the fun part. Let’s pull a model and have our first local AI conversation.

Which Model Should You Run First on Mac?

Choosing the right model for your Mac’s RAM is key. Here’s my personal recommendation based on my testing across different Apple Silicon Macs:

Run Your First Model

Open Terminal and type:

ollama run llama3.1

The first time you run this, Ollama downloads the model to your Mac. Llama 3.1 8B is about 4.7 GB — download time depends on your internet speed (typically 5–15 minutes on a normal home connection).

Once downloaded, you’ll see:

>>> Send a message (/? for help)

Type anything and press Enter. The response streams back in real time — completely offline, completely private, completely free.

>>> What are the three most important things to know about Ollama?

1. Ollama is a local AI runtime — it runs AI models entirely on your computer
2. It's free and open-source, with no usage limits or subscriptions
3. Your data never leaves your device — complete privacy by design

Getting More from Ollama on Mac

Add a ChatGPT-style Web Interface

The terminal chat works, but most people prefer a visual interface. Open WebUI is the most popular browser-based frontend for Ollama — think ChatGPT, but running entirely on your Mac.

The quickest way to install it (requires Docker Desktop for Mac):

docker run -d -p 3000:8080 \
  --add-host=host.docker.internal:host-gateway \
  -v open-webui:/app/backend/data \
  --name open-webui \
  ghcr.io/open-webui/open-webui:main

Then open http://localhost:3000 in your browser. You get a full chat interface with conversation history, model switching, and file uploads.

Don’t want to use Docker? See our guide: How to Set Up Open WebUI on Mac Without Docker →

Use Ollama from Python on Mac

If you’re a developer, Ollama’s REST API is available at http://localhost:11434. Using the official Python library:

# Install the library
pip install ollama

# Use it in Python
import ollama

response = ollama.chat(
    model='llama3.1',
    messages=[{'role': 'user', 'content': 'Explain what a neural network is in 3 sentences'}]
)
print(response['message']['content'])

Keep Ollama Updated on Mac

Ollama releases updates frequently with new model support, performance improvements, and bug fixes. To update:

• App download method: Download the latest version from ollama.com/download and replace the app in your Applications folder
• Homebrew method: Run brew upgrade ollama in Terminal
• curl method: Re-run curl -fsSL https://ollama.com/install.sh | sh — it auto-updates

Essential Ollama Commands for Mac

# See all models you've downloaded
ollama list

# Download a model without starting a chat
ollama pull mistral

# Start a chat with a specific model
ollama run deepseek-r1

# Delete a model to free up disk space
ollama rm gemma3

# Check which models are currently in memory
ollama ps

# Serve the API (starts automatically on launch, but run manually if needed)
ollama serve

# Check your Ollama version
ollama --version

Where Ollama Stores Models on Mac

Models are downloaded to: ~/.ollama/models/

To see how much space they’re using:

du -sh ~/.ollama/models/

To move models to an external drive or different location, set the OLLAMA_MODELS environment variable:

# Add to your ~/.zshrc file
export OLLAMA_MODELS="/Volumes/ExternalDrive/OllamaModels"

# Apply the change
source ~/.zshrc

Troubleshooting Ollama on Mac

macOS security warning — “Apple cannot check it for malicious software”

This is Gatekeeper — macOS’s built-in security feature. It appears for apps not downloaded from the Mac App Store. To allow Ollama:

1. Go to System Settings → Privacy & Security
2. Scroll down to the Security section
3. Look for a message saying Ollama was blocked → click “Open Anyway”
4. Confirm in the dialog that appears

This is completely safe. Ollama is open-source software — you can read every line of its code at github.com/ollama/ollama. The security warning appears because the developer hasn’t paid Apple’s notarization fee, not because the software is dangerous.

“command not found: ollama” in Terminal

If you installed via the .app method, the command-line tool needs to be on your PATH. Run:

# For Zsh (default since macOS Catalina)
echo 'export PATH="/usr/local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc

# For older Bash shells
echo 'export PATH="/usr/local/bin:$PATH"' >> ~/.bash_profile && source ~/.bash_profile

Ollama is slow on my Mac

If responses seem slow, check two things:

1. Is the model size right for your RAM? Run ollama ps — if you see “100% CPU” and you’re on Apple Silicon, Ollama may not be using the GPU. This sometimes happens when the model is too large for RAM and has to use swap memory. Solution: use a smaller model.
2. Close other heavy apps: On Macs with 8–16 GB unified memory, Google Chrome with 20 tabs open can eat 4–8 GB of RAM that Ollama needs. Quit Chrome or other heavy apps before running large models.

Ollama menu bar icon is missing

Click the Ollama app in your Applications folder to relaunch it. The menu bar icon should appear within a few seconds. If it still doesn’t appear, run ollama serve in Terminal to start the server manually.

Mac getting hot when running models

This is normal — running a large language model is computationally intensive. Apple Silicon Macs handle this well due to their efficient chip design, but the Mac Pro or MacBook Pro will spin up fans during extended model runs. MacBook Air (fanless) may thermal throttle during very long sessions with large models — this reduces speed but doesn’t cause damage.

Frequently Asked Questions

Have a question about running Ollama on your specific Mac model? Drop it in the comments — I personally test on Apple Silicon and am happy to help with specifics.

About this guide: Tested on MacBook Air M2 (16 GB), MacBook Pro M3 Pro (36 GB), and a 2019 Intel MacBook Pro (16 GB). All steps verified March 2026 using Ollama 0.6.x and macOS Sequoia 15.