MLX
Also: mlx-lm
Apple's machine-learning framework for Apple Silicon, and the model format built on it. MLX builds run models efficiently using a Mac's unified memory.
MLX is a framework from Apple for running and training models on Apple Silicon, designed around the Mac’s unified memory. Models converted to the MLX format, often published by the community as 4-bit or 8-bit quantisations, typically run faster on a Mac than the equivalent GGUF through a general engine.
For local AI on a Mac, MLX is the native path: tools like LM Studio can run MLX models directly, and the MLX-LM project handles running and fine-tuning from the command line. On other hardware it does not apply; there, GGUF and engines like llama.cpp or vLLM are the equivalent.