Meta and Apple team up for Muse Spark Open on Mac Studios

Meta and Apple are jointly backing an open-weight multimodal model with 816 billion parameters that can run on two Apple M5 Ultra Mac Studios via MLX.

What happened

Meta announced the Muse Spark Open AI model, calling it the “most capable US open-weight model to date.” The company also said it is the first model in its class that can run on your desktop using two Apple M5 Ultra Mac Studios with the latest MLX framework.

Meta and Apple, typically rivals, worked together on the launch and are publicly supporting each other. Meta CEO Mark Zuckerberg wrote on X: “Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.” Apple CEO John Ternus replied: “Amazing to run a model of Muse Spark Open’s caliber and scale this efficiently, right from your desk. Mac Studio keeps pushing the boundaries of on-device AI!”

The model is natively multimodal with 816 billion total parameters and, according to Meta, excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Meta chief AI officer Alexandr Wang told Axios the companies have been collaborating over several months. Apple also published a new technical post about MLX, Muse Spark Open, and two other open models.

Key facts

• Meta announced Muse Spark Open as the “most capable US open-weight model to date” and the first in its class to run on two Apple M5 Ultra Mac Studios with the latest MLX framework.
• Muse Spark Open is a natively multimodal model with 816 billion total parameters, with stated strengths in long-horizon tasks, complex software engineering, and multi-step reasoning.
• Meta and Apple CEOs publicly supported the launch, with Mark Zuckerberg crediting work with Apple and John Ternus praising desktop efficiency.
• Meta chief AI officer Alexandr Wang told Axios that Meta and Apple have been collaborating over several months.
• Apple published a new technical post about MLX, an array framework for ML tasks on Apple Silicon, covering Muse Spark Open and two other open models.

Our analysis

The most important shift is that a very large open-weight model is being positioned as a desktop workload on Apple silicon, not only as a cloud service. That likely matters for creators and operators who want to run private, local AI workflows without sending every prompt to a third-party API. The explicit joint support from Meta and Apple also suggests that MLX will be a first-class deployment path, which is different from open-weight releases that leave hardware support to the community.

However, the hardware requirement of two Apple M5 Ultra Mac Studios likely signals a high entry floor. This is not a model that most individual creators will run on a laptop. The practical near-term audience is likely studios, agencies, and operators who already maintain high-end Mac workstations and need local multimodal generation, long-horizon task handling, or code-heavy automation. The 816 billion total parameters figure also suggests that even on optimized Apple hardware, storage and memory requirements will likely be significant, so independent testing will matter before teams adopt it.

For social media managers and brand marketers, the release matters because it points toward future content pipelines where open-weight models can be fine-tuned or run locally for drafts, captions, image prompts, and multimodal editing. The risk is over-adoption based on a vendor announcement; the source does not report independent benchmarks, so operators should treat the launch as a signal, not a proven workflow upgrade.

What it means for operators

• Check whether your existing Mac Studio setup matches the reported requirement of two Apple M5 Ultra Mac Studios; do not assume a single lower-end machine can support Muse Spark Open as described.
• If you run high-end Apple hardware, test MLX-based local inference for private drafts, multimodal content, and long-horizon tasks before redirecting cloud AI budgets.
• Treat the Meta-Apple collaboration as a sign that Apple Silicon is a priority target for open-weight AI; follow Apple’s MLX technical post and the release notes for Muse Spark Open and the two other open models before rebuilding workflows.
• For teams without that hardware, wait for independent benchmarks and licensing clarity rather than committing content production to an unproven local setup.