MLX is Apple's open-source machine learning framework for Apple silicon Macs, with a NumPy-like Python API.
In plain words
MLX lets you build, train and run models on a Mac's own chip. It looks familiar: its core API follows NumPy and its neural-network pieces follow PyTorch, with C++, C and Swift versions too. Its big difference is unified memory: arrays live in memory shared by the CPU and GPU, so work can run on either without copying data. It computes lazily, only when results are needed.
A quick example
A developer with an M-series MacBook can load a small open model with MLX and run it locally on the laptop's GPU, then fine-tune it on their own notes without a separate graphics card.
Where you'll meet it
- Running and fine-tuning models locally on Macs.
- Machine learning research on Apple hardware.
Related
Sources
- DocsMLX README, Apple ml-explore (GitHub) · read 29 Sept 2026
- DocsMLX documentation, Apple ml-explore · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.