/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Apple's machine learning research team quietly releases MLX, an array framework to train and deploy ML models on Apple silicon, available on GitHub

Jonny Evans / Computerworld :

Computerworld Jonny Evans

Context & Ripple Effects

MLX extends Apple’s earlier positioning of its chips as machine-learning hardware: at WWDC 2023, the company emphasized ML and transformers while presenting M2 Ultra as capable of model training M2 Ultra’s model-training role.

The release creates a developer-facing layer beneath Apple’s later AI stack, from its split on-device and server model approach Apple Intelligence’s on-device and server models to direct developer access through the Foundation Models framework.

First-order effects

  • Researchers and developers can use an Apple-built array framework to train and deploy models on Apple silicon, rather than treating that hardware solely as a target for third-party tooling.
  • Publishing MLX on GitHub makes the framework accessible to the broader ML community and gives Apple a public vehicle for gathering usage and developer attention around its silicon.

Second-order effects

  • MLX can make Apple silicon more practical for ML experimentation and deployment, reinforcing demand for tooling that is tailored to a chip platform rather than hardware-agnostic by default.
  • The framework gives Apple a common technical layer that can connect hardware optimization with its eventual on-device AI developer interfaces, including the later Foundation Models access layer.

Third-order effects

  • If Apple continues to pair proprietary chips, models, and developer tooling, AI differentiation may increasingly reside in integrated stacks rather than in model access alone.
  • Open-sourcing a platform-specific framework can broaden ecosystem participation while preserving Apple’s control over the underlying hardware and operating environment.

The trend: MLX is an early example of AI platform competition shifting toward vertically integrated hardware, model, and developer-tool stacks.

Discussion

  • @luokai @luokai on threads
    Finally, Apple is making strides in open-source artificial intelligence!  They have just launched MLX, a PyTorch-style neural network framework optimized for Apple Silicon, such as MacBook models equipped with M-series chips.  MLX Source: https://github.com/... Well-documented, s…
  • @awnihannun Awni Hannun on x
    Just in time for the holidays, we are releasing some new software today from Apple machine learning research. MLX is an efficient machine learning framework specifically designed for Apple silicon (i.e. your laptop!) Code: https://github.com/... Docs: https://ml-explore.github.io…
  • @deliprao @deliprao on x
    NEWS: Apple just entered the AI open source arena by quietly releasing their new DL framework called MLX! It runs code natively on Apple Silicon with a single pip install and no other dependencies. https://github.com/... Sharing what I discovered from this initial release: [image…
  • @davewiner Dave Winer on x
    @om I'd be very suspicious of lockin and fud here, knowing apple and guessing how far behind they are. ;-)
  • @om @om on x
    Looks like Apple is doing some “AI” things to help people do AI things on their computers. This is a good development thing for all AI tinkerers :) https://om.co/...
  • @benbajarin Ben Bajarin on x
    Subtly, very interesting. Or not so subtly if you are in AI academia. It has been a challenge to get ML/LLM models to run on ANE (Apple Neural Engine). Despite @MaxWinebach's best efforts to port models to ANE.
  • @ryancarson Ryan Carson on x
    Apple is starting to show their hand in the AI/ML space. MLX, developed by Apple, is a new machine learning framework tailored for Apple silicon. It merges familiar aspects of NumPy and PyTorch with unique features like lazy computation, dynamic graph construction, and a...
  • @ylecun Yann LeCun on x
    @awnihannun Nice!
  • @awnihannun Awni Hannun on x
    The video is a Llama v1 7B model implemented in MLX and running on an M2 Ultra. More here: https://github.com/... * Train a Transformer LM or fine-tune with LoRA * Text generation with Mistral * Image generation with Stable Diffusion * Speech recognition with Whisper [video]
  • @krishnanrohit Rohit on x
    Feels a pretty big deal
  • @angeloskath Angelos Katharopoulos on x
    I am really excited about our latest work! A simple efficient framework to experiment with modern neural networks even on your laptop! 12 lines to write a transformer LM 🥳 [image]
  • @drjimfan @drjimfan on x
    This may be Apple's biggest move on open-source AI so far: MLX, a PyTorch-style NN framework optimized for Apple Silicon, e.g. laptops with M-series chips.  The release did an excellent job on designing an API familiar to the deep learning audience, and showing minimalistic examp…
  • @jsusskin Josh Susskind on x
    I'm delighted to see MLX come to life within our Apple ML research team, and I'm really excited to see how people will use and build on it now that it's been released! Lots of good things happening!