Meta has announced Muse Spark Open, an open-weight AI model intended to run locally on Apple-silicon Macs. The company says it optimized the release for Apple’s Mac Studio with the M5 Ultra chip, as first reported by TechCrunch.
The report also says Meta and Apple collaborated on the launch. Meta did not position the model as a general release for every Mac, instead naming the Mac Studio desktop and Apple’s latest high-end chip as its target hardware.
Muse Spark Open targets local AI workloads
Muse Spark Open is designed for on-device use, which puts the model’s computing work on the Mac itself. That approach differs from services that process requests only on remote servers, and it makes the capabilities of the host machine central to the experience.
Meta calls Muse Spark Open the most capable open-weight model released by a U.S. company. The statement sets a high bar for the model, though it remains the company’s own assessment rather than a result supported by independent testing included with the announcement.
The Mac focus matters because Meta has tied this release to a specific Apple desktop platform rather than treating local AI as hardware-neutral. The move also gives the Mac Studio a clearer role in the growing market for on-device models, where processor performance and available memory can determine whether a system can run a model at all.
Key access and hardware details remain unanswered
Several practical details remain unconfirmed. Meta has not provided a release date or explained how users will download or access Muse Spark Open. A reported requirement of up to 1TB of RAM raises questions about the actual hardware threshold, including whether that figure is mandatory for operation or applies only to certain configurations. The scope of any formal Apple-Meta partnership is also unspecified, and independent testing has not yet evaluated Meta’s capability claim.
This article was produced with AI assistance from multi-source reporting and is published under our editorial standards.
