Models·
Meta and Apple Join Forces to Boost Muse Spark Open Performance on Mac Studio
The two tech giants unveil an open‑weight multimodal model that runs efficiently on dual M5 Ultra Mac Studios via Apple’s MLX framework, marking a rare collaboration between rivals.
Meta announced Muse Spark Open, an open‑weight multimodal model with 816 billion parameters designed for long‑horizon reasoning, complex software engineering and multi‑step tasks.
The model is optimized to run on two Apple M5 Ultra Mac Studios using the newly updated MLX array framework, which enables efficient tensor operations on Apple Silicon without relying on discrete GPUs.
Both companies highlighted the partnership, with Meta’s CEO noting the model’s performance on Mac Studio and Apple’s CEO praising the ability to run such a large model locally.
Why it matters for GPU / AI infrastructure
This demonstration shows that frontier‑scale models can be executed on‑device using unified memory architectures, offering a preview of how inference workloads might shift toward power‑efficient silicon and influencing future GPU cloud strategies for low‑latency, edge‑focused AI services.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- muse spark
- apple silicon
- mlx
By AiGpu Editorial · Editorial rewrite based on public reporting (The Verge AI)
← All articles