Running on a Mac
What is MLX, and why does it matter on a Mac?
MLX is Apple's open-source machine learning framework built for Apple silicon. Its design matches the hardware: arrays live in shared memory, so the same data is usable by the CPU or GPU without copying. Several popular local-AI apps now run MLX models under the hood.
The unified-memory advantage only pays off if the software is written to use it. MLX is Apple’s answer: an open-source “array framework for machine learning on Apple silicon, brought to you by Apple machine learning research.”[1] An array framework is the layer that does the math underneath a model, similar in spirit to NumPy or PyTorch, but built for this hardware.
What makes MLX fit the Mac is how it handles memory. In its design, “Arrays in MLX live in shared memory. Operations on MLX arrays can be performed on any of the supported device types without transferring data.”[1] In plain terms, the same data is usable by the CPU or the GPU with no copy step, which is exactly the shape of Apple’s hardware. MLX offers programming interfaces in Python, C++, C, and Swift, with the Python side following NumPy conventions so it feels familiar.[2]
You do not have to write code to benefit from it. Several friendly apps for running models locally on a Mac can now use MLX models. The page on getting started covers picking one.
Ways to run MLX models on a Mac
- MLX (mlx-lm) ↗
Apple's own command-line tools and Python package for generating text and fine-tuning models on Apple silicon.
- Ollama ↗
A simple way to download and run open models locally, on Mac and other platforms.
- LM Studio ↗
A desktop app for running open models privately on your own computer, with Apple MLX support.
References
- MLX: An array framework for Apple silicon — Apple (ml-explore on GitHub)
- MLX project page — Apple Open Source