AI2 mins read

Meta launches Muse Spark Open, an open-weight AI model built to run on Macs

Meta and Apple have teamed up on Muse Spark Open, a new open-weight AI model designed to run locally on Apple silicon and aimed at coding, agentic workflows, and multimodal tasks.

Meta’s new model is built for local Mac use

Meta has launched Muse Spark Open, an open-weight AI model it describes as its most capable open-weight model in the United States. The model was designed to work with Macs and run locally on Apple’s computers, tying Meta’s latest AI push directly to Apple silicon.

For readers, the key takeaway is practical: this is not just a cloud-only model announcement. Meta is positioning Muse Spark Open as a desktop-capable AI system for users with sufficiently powerful hardware.

The specs point to serious ambition — and serious hardware needs

Muse Spark Open uses 816 billion parameters and a 1 million token context window, according to the article. It is natively multimodal and built for complex coding tasks and agentic labor.

The local-use pitch comes with a major caveat: the model requires at least 1 terabyte of RAM to run locally. TechCrunch reports it will run best through a Mac Studio equipped with the M5 Ultra chip.

Coding and creative workflows are the early focus

Meta paired Muse Spark Open with Muse Code, its AI coding terminal, and used it across agentic workflows. One example cited involved coding fixes for a Mastodon client, including bug fixes, new features, and a full build.

The company also deployed the model with DaVinci Resolve, where it was able to review and crop footage. That signals Meta is presenting the model as useful beyond chat, with coding, desktop automation, and creative software tasks in view.

Why open-weight matters here

Open-weight models can offer cost efficiency, transparency, and flexibility that closed proprietary models do not provide. For developers and companies, that can make experimentation and customization more accessible, depending on hardware and deployment needs.

Meta is making Muse Spark Open available now through multiple API platforms. The immediate question for users is whether they want local control badly enough to meet the model’s steep hardware requirements.

Discover More