Apple Releases 20 New Open Source AI Models

New open source models are optimized for running language and diffusion models on Apple devices

Ben Wodecki, Jr. Editor

June 20, 2024

2 Min Read
The Apple logo above a new store in Kuala Lumpur, Malaysia
Annice Lyn/Getty Images

Hot off the heels of its OpenAI integration, Apple has expanded its support for open-source AI, making 20 new AI models available to developers.

Apple previously only open-sourced one AI model, OpenELM, which launched in April.

The iPhone maker has now quietly published 20 new models to its Hugging Face page, under its CoreML framework to power on-device AI workloads.

CoreML comes in several versions, including DETR-Resnet50 for semantic segmentation and FastViT for image classification. The models support domain-specific workloads, including images, processing text and converting audio to text.

The new models are all optimized for running generative AI and machine learning applications on Apple’s hardware, allowing for models to run without a network connection.

Apple said the CoreML models enable open source developers to flexibly deploy large language models and diffusion models on their devices while reducing memory footprint and power consumption.

CoreML users can convert third-party training libraries such as TensorFlow and PyTorch, enabling developers to integrate pre-trained models into Apple’s ecosystem.

Vaibhav Srivastav, a Hugging Face data scientist, utilized one of the new CoreML models to enhance the speed of image generation for the new Stable Diffusion 3.

Related:Apple Integrates ChatGPT Across Platforms, Unveils Apple Intelligence

The CoreML models work on Apple Mac devices running macOS High Sierra and iPhones on iOS 11 and above.

Clement Delangue, Hugging Face’s cofounder and CEO revealed earlier this month that the open source platform is actively working with Apple to grow the community around its MLX framework for machine learning.

MLX was Apple’s first foray into open source AI, providing developers with tools to train and run large language models on its hardware.

At the company’s recent Worldwide Developer Conference, Apple expressed support for running AI features on-device, rather than conducting the processing in the cloud, a position Hugging Face’s CEO also supported.
Not everyone is in favor of the idea, however, with Elon Musk threatening to ban staff at his companies from using Apple devices, describing the integration as “creepy spyware.”

Read more about:

ChatGPT / Generative AI

About the Author

Ben Wodecki

Jr. Editor

Ben Wodecki is the Jr. Editor of AI Business, covering a wide range of AI content. Ben joined the team in March 2021 as assistant editor and was promoted to Jr. Editor. He has written for The New Statesman, Intellectual Property Magazine, and The Telegraph India, among others. He holds an MSc in Digital Journalism from Middlesex University.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like