Local LLM runnersMITPython

MLX LM

Run and fine-tune LLMs on Apple Silicon with Apple’s MLX framework.

Stars
7.3K
Forks
1.1K
Open issues
236
Last commit
2026-10-10

Live from the GitHub API · updated 2026-10-15

What is MLX LM?

MLX LM uses Apple’s MLX array framework to run LLMs efficiently on M-series Macs, with quantization, LoRA fine-tuning and direct access to thousands of mlx-community models on Hugging Face.

Key features

  • Optimized for Apple Silicon unified memory
  • Quantize and convert Hugging Face models
  • LoRA and full fine-tuning
  • OpenAI-compatible server

System requirements

Platforms
macOS
GPU
Apple Silicon (M1 or newer)
Python
3.9+

How to install MLX LM

Summarized from the project’s README. Projects change quickly — check the GitHub page for the latest instructions.

pip

  1. 1
    Install.
    pip install mlx-lm
  2. 2
    Generate text.
    mlx_lm.generate --model mlx-community/Llama-3.2-3B-Instruct-4bit --prompt "Write a haiku about the sea"

Strengths

  • Fastest option on Macs
  • Easy fine-tuning

Limitations

  • Apple Silicon only

MLX LM alternatives compared

SoftwareStarsLicensePlatformsVRAM
MLX LM

Run and fine-tune LLMs on Apple Silicon with Apple’s MLX framework.

7.3KMIT—
Ollama

Get up and running with large language models locally.

183.2KMIT—
llama.cpp

LLM inference in C/C++ — the engine behind most local AI apps.

131.2KMIT—
Jan

Open-source ChatGPT-style desktop app that runs 100% offline.

45KAGPL-3.0—
GPT4All

Private, on-device chat with local document Q&A.

77.4KMIT—
LocalAI

Self-hosted, OpenAI-compatible API for text, images, audio and more.

49.5KMIT—

FAQ

Is MLX LM free?

Yes — MLX LM is open-source software released under the MIT license. Check the license (and any model weight licenses) before commercial use.

Does MLX LM run on Windows, Mac and Linux?

MLX LM supports: macOS.

What hardware do I need for MLX LM?

Apple Silicon (M1 or newer)

How do I install MLX LM?

1. Install. 2. Generate text.

What are the best MLX LM alternatives?

Popular alternatives include Ollama, llama.cpp, Jan, GPT4All. See the comparison table on this page.