Official
Intermediate
Official foundation
MLX
The core MLX framework for arrays, neural networks, autograd, and Apple silicon-optimized ML work.
Why it matters: This is the root layer. If you want to understand the ecosystem instead of just using it blindly, start here.
Who it's for: Builders who want the actual foundation, not just wrappers around it.
Try next: Read the docs lightly, then jump to MLX Examples or MLX LM so the abstractions become real.
Quick note: Foundational, not the fastest first win.
Official
Beginner-friendly
LLM lane
MLX LM
The most practical entry point for running and fine-tuning language models in the MLX ecosystem.
Why it matters: If you want one fast, useful first win with MLX, this is usually the move.
Who it's for: People who want local LLMs on a Mac without building a full stack from scratch.
Try next: Pair it with compatible weights from mlx-community or the Hugging Face MLX browser.
Quick note: Best first lane for most newcomers.
Official
Beginner-friendly
Docs + community
MLX Examples
A repo of example projects and demos showing how MLX gets used in practice.
Why it matters: Examples are where the theory stops being vague.
Who it's for: People who learn faster from concrete demos than from abstract docs.
Try next: Pick one example that feels close to your goal and get it running end to end.
Quick note: Great bridge between docs and real use.
Official
Beginner-friendly
Docs + community
Official MLX docs
The documentation hub for installation, APIs, and core MLX concepts.
Why it matters: When you want the cleanest source of truth, this is it.
Who it's for: Anyone who wants the official explanation before community shortcuts.
Try next: Read just enough to orient yourself, then switch to examples or MLX LM for hands-on learning.
Quick note: Best used in small doses, not as your only learning path.
Community
Beginner-friendly
Model discovery
mlx-community on Hugging Face
A major source of MLX-compatible model weights so you can try models without conversion drama.
Why it matters: This is where practical experimentation becomes plug-and-play.
Who it's for: People who want to run models fast instead of wrestling with formats first.
Try next: Pick a model that matches your use case, then run it through MLX LM.
Quick note: Huge practical unlock for newcomers.
Community
Intermediate
Multimodal + media
mlx-vlm
A community project for running vision-language models with MLX.
Why it matters: It proves MLX is not just about text chat models.
Who it's for: People curious about image + text workflows, not just pure LLM use cases.
Try next: Treat it as a community frontier project after you've done one simpler MLX lane first.
Quick note: Exciting, but less foundational than the official repos.
Community
Beginner-friendly
Multimodal + media
MFLUX
MLX-native implementations of modern image generation models on Apple Silicon.
Why it matters: This is one of the clearest ways to show newcomers that MLX can drive practical image workflows too.
Who it's for: People who want a visual, tangible MLX project instead of another abstract repo.
Try next: Try it after you understand the core stack so the image lane feels connected instead of random.
Quick note: Strong demo value and very shareable.