#1111th of 15 in ML frameworks and libraries
mlx
NumPy-style array framework for machine learning on Apple silicon
- Stars
- 28.8k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Sep 2026
- Language
- C++
24,527 on 24 Mar 202628,774 on 13 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
MLX is an array framework from Apple machine learning research with Python, C++, C and Swift APIs. It builds lazy, dynamic computation graphs, offers composable transforms for autodiff and vectorization, and includes PyTorch-like mlx.nn and mlx.optimizers packages. Arrays live in unified memory, so CPU and GPU ops share data without copies. Linux CUDA and CPU-only packages exist.
Who it is for: ML researchers and engineers training or running models on Macs
Strengths
- Python API follows NumPy; mlx.nn and mlx.optimizers follow PyTorch
- Unified memory: CPU and GPU ops share arrays without transfers
- Composable transforms for autodiff, vectorization and graph optimization
- Python, C++, C and Swift APIs; installs via pip on macOS and Linux
Weaknesses
- Designed for Apple silicon; Linux CUDA and CPU backends are add-on installs
- Example models live in a separate mlx-examples repo
- Supported devices are only CPU and GPU per the README
- Model zoo and ecosystem size compared to PyTorch not stated in README
What it needs
- GPU optional
- Models: LLaMA, Stable Diffusion, Whisper, LoRA fine-tuning
Also in ML frameworks and libraries
See all 15| Rank | Project | Score |
|---|---|---|
| 1 | transformersModel-definition library for text, vision, audio and multimodal models | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 92 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 86 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 82 out of 100 |
| #5 | tensorflowTensor and neural network framework with Python and C++ APIs | 80 out of 100 |