Ollama in Infinity.

Free models that run on your Mac. Infinity installs and starts Ollama for you.

What you need

Nothing but disk space. Small models need about 2 to 5 GB; 8 GB of memory or more is best.

Open Infinity. On first run, choose Ollama when Infinity asks you to pick your models. Later, go to Settings ▸ Accounts and models ▸ Ollama.
Press Set up Ollama. Infinity downloads and starts it quietly; it takes about a minute. Already have Ollama? Infinity uses the one you have.
Open Settings ▸ Models by job and pick a chat model sized for your Mac, such as Llama 3.2 3B or Qwen 3 4B. Infinity downloads it.
Start a chat. Everything stays on your Mac, and it works without the internet.

Infinity Code and assistants need a bigger context window than Ollama’s default. Infinity offers to raise it to 64K, which restarts Ollama.
Some features need a newer Ollama (0.35 or later). Infinity tells you and links to the update.
Ollama runs on Apple silicon and Intel Macs. Big models are much faster on Apple silicon.