Meta and Apple partner to improve Muse AI performance on Macs
apple meta
| Source: The Verge | Original article
Meta and Apple partner to run the new Muse Spark open‑weight AI model on desktop Macs using Apple M5 Ultra Mac Studios and the MLX framework.
Meta and Apple announced a joint effort to bring Meta’s newly unveiled Muse Spark Open model to high‑end Mac Studio machines. Described by Meta as “the most capable US open‑weight model to date,” Muse Spark Open is also billed as the first model of its class that can run locally on a desktop, specifically on two Apple M5 Ultra Mac Studios using the latest version of Apple’s MLX machine‑learning framework.
The partnership marks a shift from the cloud‑centric deployment that has dominated large language models. By enabling developers to run a massive open‑weight model on local hardware, the collaboration promises lower latency, reduced data‑transfer costs, and greater control over proprietary workloads. Apple’s update to MLX underscores its commitment to supporting third‑party AI workloads on its silicon, while Meta gains a showcase for its model on a premium platform without relying on external data‑center resources.
As we reported on 16 October, Meta had already launched an open‑weight AI model that could run on Macs. Today’s announcement adds a new, more capable model and a tighter integration with Apple’s software stack, signaling a deepening of the two companies’ cooperation after a period of rivalry.
Going forward, observers will watch for performance benchmarks that compare Muse Spark Open on Mac Studio against cloud alternatives, and for any signals that Apple might embed Meta’s models into its broader Apple Intelligence suite. Further hardware support—such as compatibility with the upcoming Mac Book Pro line—or expanded licensing terms for developers could determine how widely the local‑run model is adopted across the AI ecosystem.
Sources
Back to AIPULSEN