Source: techcrunch.com/2026/10/15/meta-launches-an-open-weight-model-that-runs-on-macs
Muse Spark Open has 816 billion parameters and a 1 million token context window. Meta says its evaluation results show better performance than other popular open-weight models. The model is natively multimodal and aimed at complex coding tasks and agentic labor.
Meta says the model is designed to run locally on Apple's computers, but Engadget reports that it needs at least 1 terabyte of RAM. The Mac Studio with the M5 Ultra chip is the configuration expected to run it best. Meta is making the model available now through a number of API platforms.
Meta paired the model with Muse Code, its AI coding terminal, and deployed it in agentic workflows. Those tests included coding fixes for a Mastodon client, covering bug fixes, new features and a full build. Meta also reports that the model reviewed and cropped footage in DaVinci Resolve.
Why it matters if you build agents
Developers building local coding or desktop agents on Apple hardware now have another open-weight model to evaluate, though the memory requirements are steep. Teams that want more cost efficiency, transparency or control than closed models offer get a new candidate to test. Benchmark claims from Meta should be checked against your own workloads before you rely on them.
Get stories like this every weekday
One short email with the day's agent tooling news.