Run a Local LLM with OpenClaw on Your Mac Mini

This guide provides a step-by-step tutorial for setting up a local Large Language Model on a Mac Mini to power OpenClaw agents, eliminating the need for paid APIs. It recommends using llama.cpp with Metal acceleration for optimal performance, bypassing the more common Ollama setup. The process involves building llama.cpp from source, downloading a specific quantized model like Qwen 3.5-9B, and using a compatible agent template. The tutorial also covers running the model as a persistent service using llama-server and macOS's launchd, and concludes with reconfiguring OpenClaw to use the local model endpoint.