Meta launches new Muse Spark Open model, optimized to run on Mac Studio

Meta released the weights for Muse Spark Open, which it calls the most capable US open-weight model to date, a natively multimodal model with 816B total parameters (42B active) and a 1M-token context window, available now through Hugging Face. Meta said Muse Spark Open is optimized to run on the Mac Studio with M5 Ultra using Apple's latest MLX, whose unified memory architecture lets the M5 Ultra's integrated GPU use up to 512 GB per Mac Studio and, at 1.2 TB/s of memory bandwidth, allows two Mac Studios to hold an entire model. Meta tested on-device agentic workflows with Muse Code powered by Muse Spark Open, including shipping real iOS changes to the open-source Mastodon client Ice Cubes and driving DaVinci Resolve end to end.

Meta today unveiled https://x.com/finkd/status/2110778025629347999?s=20 Muse Spark Open, which it says is the “most capable US open-weight model to date.” For Apple users, Meta says Muse Spark Open is optimized to run on the Mac Studio powered by M5 Ultra. In a blog post https://research.meta.ai/blog/muse-spark-open , Meta explains that Muse Spark Open is natively multimodal and brings “leading agentic capabilities to your desktop for the first time.” Today, we’re releasing the weights for Muse Spark Open, the most capable US open-weight model to date. Muse Spark Open is a natively multimodal model with 816B total parameters 42B active and a 1M-token context window. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Meta’s blog post specifically touts that Muse Spark Open leverages Apple’s latest version of MLX. MLX is an open source array framework https://opensource.apple.com/projects/mlx/ created by Apple and optimized for the unified memory architecture of Apple silicon. Using the latest version of MLX, Meta says that Muse Spark Open was optimized for Mac Studio with M5 Ultra, and that “together they’re the best way to run a powerful model on your own desktop.” Apple’s unified memory architecture lets the M5 Ultra’s integrated GPU use up to 512 GB per Mac Studio, so two machines can hold an entire model. And the chip’s 1.2 TB/s of memory bandwidth is fast enough to read the small slice each token needs. The result is two Mac Studios behave like one supercomputer with a terabyte of unified memory. We tested a range of agentic workflows running entirely on-device, using Muse Code powered by Muse Spark Open, including shipping real iOS changes to Ice Cubes an open-source Mastodon client — bug fixes, new features, full build — and driving DaVinci Resolve end to end, with the model reviewing footage frame by frame to find the strongest moments and crop them vertically. Muse Spark Open is available now through Hugging Face https://huggingface.co/collections/meta-models/muse-spark-open . Apple has also published a new article on its Machine Learning Research website https://machinelearning.apple.com/research/mlx-distributed-inference touting how MLX on Mac “lets AI researchers and developers experiment locally with the latest models and methods in a private environment.” Follow Chance : Threads https://www.threads.com/@chancehmiller , X http://x.com/chancehmiller , and Instagram http://instagram.com/chancehmiller FTC: We use income earning auto affiliate links. More. https://9to5mac.com/about/ affiliate our homepage http://9to5mac.com/ for all the latest news, and follow 9to5Mac on exclusive stories https://9to5mac.com/feature/exclusive/ , reviews https://9to5mac.com/guides/review/ , how-tos https://9to5mac.com/guides/how-to/ , and subscribe to our YouTube channel https://www.youtube.com/9to5mac