~/runtime/mlx
MLX

MLX

Apple's array framework for Apple Silicon. The substrate beneath oMLX and LM Studio MLX.

License
MIT
Platform
macOS · Apple Silicon
Model formats
MLX
API
Python (mlx-lm, mlx-vlm) · Swift · C++ · C

What it is.

$ ./vrambudget --runtime mlx

Apple's machine learning research framework, designed for Apple Silicon's unified memory architecture. NumPy-style Python API; full C++, C, and Swift bindings. Lazy evaluation, dynamic graphs, composable function transformations. Not a serving runtime on its own; you reach for `mlx-lm` (LLMs) or `mlx-vlm` (vision-language) or build on top with oMLX. Apache 2.0, actively developed by Apple ML Research.

Install.

$ pkg install mlx
pip install mlx
pip install mlx[cuda] # Linux NVIDIA backend

Supported platforms: macOS (Apple Silicon), Linux (CUDA), Linux (CPU)

Features.

$ cat features.md
Unified memory

Arrays live in shared memory; ops run on CPU or GPU without copying data. Native fit for Apple Silicon's memory architecture.

Familiar APIs

NumPy-style Python. PyTorch-style `mlx.nn` and `mlx.optimizers` for building models. Cross-language: Swift, C++, C.

Lazy + dynamic

Computations materialize only when needed. Dynamic graph construction; no slow recompiles on shape changes.

mlx-lm + mlx-vlm

Official LLM and vision-language model packages. Where you actually serve text and multimodal models from Python.

Function transforms

Automatic differentiation, vectorization, computation graph optimization. Composable, JAX-style.

Best for
  • ▸Apple Silicon inference (the unified-memory advantage)
  • ▸Researchers extending the framework with custom ops
  • ▸Building higher-level tools (oMLX is built on this)
  • ▸Training and fine-tuning on Apple Silicon (LoRA on a MacBook)
Caveats
  • ▸Not a server. You wrap it in mlx-lm, oMLX, or your own code
  • ▸Apple Silicon native; Linux paths exist but are not the primary target
  • ▸The model zoo lives at huggingface.co/mlx-community (community-maintained)

Links.

$ ls -1 ./external
↗ homepagehttps://github.com/ml-explore/mlx↗ githubhttps://github.com/ml-explore/mlx↗ docshttps://ml-explore.github.io/mlx/build/html/index.html

Compare to…

$ ./vrambudget --compare-runtimes
$ altOllamacross-platform · climacOS · Linux · Windows · Docker$ altLM Studiocross-platform · guimacOS · Linux · Windows$ altvLLMserver · productionNVIDIA CUDA (server)$ altoMLXapple · agenticmacOS Sequoia · M1 / M2 / M3 / M4 / M5

Discussion.

$ gh discussion list

// sign in with github to leave a comment. threads live in the repo's discussions tab.