Ollama <-> Tailscale <-> macOS

This guide will walk you through setting up Ollama on your Mac and configuring it to be accessible across your network using Tailscale.

Prerequisites

First, you need to install Tailscale: https://www.tailscale.com

Tailscale

Tailscale serves as a secure, zero-config mesh VPN.

In this context, its purpose is to create a private, encrypted network between your Mac (where Ollama is running) and other machines you use (clients).

This allows you to securely access the Ollama API, which hosts your Large Language Models (LLMs) from any location on the internet, as if those other machines were on the same local network.

It bridges the gap, enabling remote devices to communicate directly with your local Ollama service.

Installation Steps

1. Install Ollama: Once you have created an account and installed Tailscale, install Ollama on your Mac using Homebrew:
   $
2. Download a Model: Install a model of your preference before proceeding. You can check the available models here: https://ollama.com/search

Configuring the Environment Variable

Before starting Ollama as an API service via Homebrew, you must set the environment variable OLLAMA_HOST to 0.0.0.0.

This adjustment is necessary so that other machines connected via Tailscale can access the Ollama API. Without this change, other machines will not be able to connect to the Ollama API.

1. Create the Launch Agent File

Create the file ~/Library/LaunchAgents/setenv.ollama.plist with the following content:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
 "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
  <dict>
    <key>Label</key>
    <string>setenv.ollama</string>

    <key>ProgramArguments</key>
    <array>
      <string>launchctl</string>
      <string>setenv</string>
      <string>OLLAMA_HOST</string>
      <string>0.0.0.0</string>
    </array>

    <key>RunAtLoad</key>
    <true/>
  </dict>
</plist>

2. Load the Configuration

Load the new file using launchctl:

3. Start the Ollama Service

You can now start Ollama as an API service:

Testing Connectivity

After the service has started, on another machine with Tailscale configured on the same account, perform a test connection:

Instructions for Testing:

• Replace {HOST} with the host provided by Tailscale.
• Replace {ID} with the generated ID.
• Replace {MODEL} with the model you installed via Ollama.