Meta and Apple make AI models run locally on Macs

Meta and Apple have launched Muse Spark Open, an AI model designed to run directly on Apple silicon, according to TechCrunch and The Verge. This partnership means businesses can now use advanced AI on Mac Studio computers without relying on cloud services.

Why it matters

For companies using AI, this launch brings more control. According to Meta, Muse Spark Open is the "most capable US open-weight model to date" and can run with two Apple M5 Ultra Mac Studios using Apple's MLX framework. Running models locally gives businesses options for handling sensitive data, and can reduce ongoing cloud costs. It also means less dependence on external providers for day-to-day AI workloads.

Apple and Meta both say this is the first time a model of this class can run on Macs, moving some high-end AI work away from large data centers. The Verge reports that the companies have worked together to make it possible to cluster Mac Studios for these workloads.

How Muse Spark Open works on Macs

Meta says Muse Spark Open is an "open-weight" model. This means companies can download and run it on their own hardware, rather than just accessing it through an API or web interface. The model is optimized for Apple's M5 Ultra chips, using the MLX framework. According to The Verge, this setup lets users cluster multiple Mac Studios for heavier jobs. The companies say this is aimed at developers and operators who want full control over their AI systems, including privacy and security.

CNBC reports that Apple and Meta's collaboration focuses on making Muse Spark Open easier to run on Mac Studios, especially for organizations that need to keep data on-site. The model does not require a constant internet connection and does not send data back to Meta or Apple during operation. This approach is different from cloud-based AI tools, which often require sending business data to external servers. For comparison, other recent AI launches, such as OpenAI's GPT-6 Astra, still rely mostly on remote infrastructure.

What this means if you run a business

For business owners and operators, this launch offers a new way to deploy advanced AI tools. If your company already uses Mac Studios, you can now run a top-tier AI model locally, keeping customer or operational data inside your own network. This can help with compliance and reduce risks around data sharing. It could also lower your total cost if you are currently paying for cloud-based AI processing or storage.

The move also signals a shift. Tech companies are starting to build AI models that run on standard office hardware, not just in massive data centers. This gives you options when negotiating with vendors, since you are less locked into a single cloud provider. If you are in a regulated industry, or deal with sensitive information, the ability to run AI in-house is a practical advantage. However, the model is optimized for the latest Apple hardware and MLX framework, so older Macs or Windows systems are not supported for this use case.

For now, the main benefit is control: you decide how and where your data is processed, and you set the security standards. As more AI models become available for local deployment, expect pressure on cloud AI pricing and new demands on IT teams to manage these systems in-house.