mlx-lm ☆

« Back to VersTracker

Description:
Run LLMs with MLX

Type: Formula | Latest Version: 0.29.0@0 | Tracked Since: Dec 16, 2025

Links: Homepage | GitHub | formulae.brew.sh

Stars: 3,153 | Forks: 336 | Language: Python | Category: Ai ml

Tags: llm machine-learning apple-silicon python mlx

Install: brew install mlx-lm

About:
mlx-lm is a Python library that enables efficient inference and fine-tuning of Large Language Models (LLMs) on Apple silicon using the MLX framework. It provides a simple interface to run popular open-source models locally, leveraging the performance and memory efficiency of Apple's hardware. This allows developers and researchers to experiment with LLMs without requiring specialized GPU setups.

Key Features:

Run popular LLMs locally on Apple silicon
Efficient inference using the MLX framework
Simple Python API for loading and generating text
Support for model fine-tuning
Optimized for memory and performance on macOS

Use Cases:

Local LLM inference for privacy-sensitive applications
Prototyping and experimenting with open-source language models on Mac
Educational and research purposes for understanding LLM behavior

Alternatives:

llama.cpp – Also runs LLMs locally on CPU/Apple Silicon, but mlx-lm is specifically built on the MLX framework for deeper Apple hardware integration.
transformers (by Hugging Face) – A more general library for transformers models; mlx-lm is a specialized, optimized backend for Apple silicon via MLX.

Version History

Detected	Version	Rev	Change	Commit
Dec 19, 2025 10:04am	0.29.0	0	VERSION_BUMP	98344d65
Dec 16, 2025 6:41pm		0	VERSION_BUMP	7dfcc68f
Dec 4, 2025 4:34pm		0	VERSION_BUMP	b8999b14
Dec 4, 2025 10:37am		0	VERSION_BUMP	48391351
Oct 15, 2025 8:09pm		1	VERSION_BUMP	19194ddf
Sep 30, 2025 5:13pm		0	VERSION_BUMP	b67a41cd