Apple and Meta Partner to Bring Meta's Open AI Model to Mac Studio
Apple and Meta have announced a partnership allowing developers to run Meta's Muse Spark Open AI model locally on Mac Studio desktops equipped with M5 Ultra chips.
After years of public rivalry and fierce disputes, Apple and Meta are joining forces in the artificial intelligence sector. The two technology giants announced a partnership that will allow developers to run Meta's new artificial intelligence model, Muse Spark Open, directly on Apple's high-performance Mac Studio desktop computers without relying on cloud services.
This marks a significant shift in relations between the companies, which were long characterized by mutual suspicion and clashes over user privacy and digital advertising markets. According to a CNBC report, the collaboration will enable developers to connect two to four Mac Studio computers to run the model for coding, video analysis, and complex tasks using AI agents. Operating locally rather than on remote servers provides an added layer of security while eliminating cloud usage costs for users.
Technical Specifications and Local Execution
Muse Spark Open is based on a Mixture of Experts architecture, which activates only relevant parts of the model for each task rather than utilizing the entire system. It is an open-weight model, meaning developers can download its parameters and run it on their own hardware subject to limited commercial use licenses.
Alexander Wong, head of AI at Meta, explained that many powerful open AI models today are too large to run on standard personal computers. However, the massive memory capacity offered by Mac Studio computers makes running Muse Spark Open practically feasible. "We worked with Apple to bring advanced AI agent capabilities to the desktop for the first time, powered by Mac Studio computers equipped with M5 Ultra chips," Wong said.
Developers can download a customized version of the model from platforms like Hugging Face and execute it on Apple silicon without requiring additional modifications. The company also published new benchmark performance metrics and technical documentation demonstrating how its MLX development framework enables splitting model execution across multiple computers.
High Hardware Costs
However, the solution comes with a steep price tag. A Mac Studio featuring an M5 Ultra chip and 256 gigabytes of unified memory starts at $10,799, and running the model requires at least two computers—representing an investment exceeding $21,000. Apple is also scheduled to launch a higher-end version with 512 gigabytes of memory later this month.
The partnership highlights a turning point after years of tension that peaked in 2011 when Apple introduced privacy policy changes restricting ad-tracking capabilities on iPhones. Despite past friction and criticism between former Apple CEO Tim Cook and Meta CEO Mark Zuckerberg, recent leadership changes and strategic realignments have paved the way for technical cooperation, offering developers powerful local AI tools while strengthening Apple's hardware positioning in the artificial intelligence landscape.
