Llama

Llama is a macOS menu bar app for running local LLMs.

How it works

When you start Llama, it runs a local server at http://localhost:9931/v1.

If you have llama.cpp installed, Llama uses it. Otherwise, it installs a prebuilt binary for your Mac. Models you've already installed via llama.cpp show up in the app automatically. You can install any GGUF model from Hugging Face, and Llama also recommends models that fit your Mac's hardware.

You can chat with any model in the built-in WebUI, connect other apps (coding agents, chat UIs, editors), or use the API directly. Models load when requested and unload when idle, so they don't take up memory when not in use.

Features

• 100% local — Models run on your Mac; no data ever leaves it
• Small footprint — 4 MB native macOS app
• Zero configuration — models are auto-configured with optimal settings for your Mac
• Model recommendations — a built-in list of models your Mac can run, installable in one click
• Standard storage — models live in the Hugging Face cache, shared with llama.cpp and other tools
• Built on llama.cpp — from the GGML org, developed alongside llama.cpp

Example requests

List installed models:

Send a message to a model:

See complete API reference in the llama.cpp server docs (opens in a new tab).

Custom model settings

The app configures each model for your Mac and writes that config to models.ini, which it regenerates on every launch. To change a setting, or add one the app doesn't set, edit ~/.config/llama/models.user.ini instead -- the app only reads that file, and merges it into the config it generates.

Use the same section header the app uses (org/repo:QUANT), and list only the keys you want to change. Keys are llama serve options without the leading dashes.

A section for a model the app didn't find is passed through as-is, which is how you point at weights or a draft model from another repo.

Anything you set here wins over the app's own value, including settings derived from how much memory your Mac has. If a key can't be applied, the app ignores the file and the menu says which option was at fault.