Meta launches Muse Spark Open for local Mac deployment
Meta released Muse Spark Open, an 816-billion-parameter model designed for local execution on high-memory Macs. The release targets complex coding and agentic tasks with a one-million-token context window.
Meta has launched Muse Spark Open, which the company describes as the most capable open-weight model in the United States. The model is specifically engineered to run locally on Apple hardware, requiring at least 1 terabyte of RAM for operation. Meta states that optimal performance is achieved on the Mac Studio equipped with the M5 Ultra chip. The architecture features 816 billion parameters and supports a context window of one million tokens, positioning it for complex coding tasks and agentic labor. Meta claims the model outperforms other popular open-weight alternatives based on evaluation results released on Thursday. To demonstrate its capabilities, Meta paired the model with Muse Code, its AI coding terminal, and deployed it in agentic workflows. These included fixing bugs and adding features to a Mastodon client, as well as reviewing tasks in DaVinci Resolve video editing software. The model is natively multimodal and is available immediately through various API platforms.