Meta and Apple team up to make Muse AI models run better on Macs
Meta says Muse Spark Open can run on a desktop setup of two Apple M5 Ultra Mac Studios, following several months of collaboration with Apple. The partnership between the usually rival companies highlights Apple hardware and its MLX software framework as a platform for running large AI models locally.
Meta describes Muse Spark Open as a natively multimodal model with 816 billion parameters, suited to long-horizon tasks, software engineering and multi-step reasoning. The company calls it the most capable US open-weight model to date and says its weights are being released; Apple also published technical material on MLX and the model.
- Meta and Apple collaborated to run Muse Spark Open on Mac hardware.
- The desktop setup uses two M5 Ultra Mac Studios.
- The model has 816 billion parameters.
New here? Start with this
Meta and Apple are large technology companies that often compete in areas such as devices, software and online services. Meta develops artificial intelligence models, while Apple makes computers and software that can run AI tasks on the device rather than sending them to remote servers.
AI models are trained using large collections of data and can be designed to handle different kinds of input, such as text and images. “Open-weight” means a model’s learned parameters are made available for others to download and run, though this does not necessarily make every part of its development or training data public.
Running a very large model locally can require substantial computing power, so support for Apple’s hardware and MLX software could make its computers more useful to AI developers. The collaboration also shows how companies that compete in some markets may work together on the tools needed to run AI systems.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The collaboration could make powerful AI more accessible by showing that a large open-weight model can run locally on commercially available Mac hardware. Supporters may value the potential for greater user control over data, wider experimentation and a stronger software and hardware ecosystem around Apple’s MLX framework.
The case against
The demonstration may say more about a specialised, expensive two-computer setup than about practical access for most users, and claims about the model’s capabilities come from the companies involved. Critics may also argue that promoting local execution does not by itself resolve concerns about the model’s safety, reliability or real-world usefulness.
AI Americas Business Companies Geopolitics Politics Technology World