--- title: "Meta Releases Weights of Muse Spark Open AI Model, Turning Mac Studio Pair into a Supercomputer" description: "Meta has released the weights of the multimodal Muse Spark Open AI model with 816B parameters, optimized for local execution on desktop Macs with M5 Ultra chips via the MLX framework." date: 2026-10-16T08:52:53.910Z lang: en url: https://xab.info/en/posts/meta-releases-muse-spark-open-ai-for-mac tags: [meta, ai, muse-spark-open, apple-silicon, mac-studio, open-source, mlx] author: "Alexander Buran" publisher: "XAB.info" --- # Meta Releases Weights of Muse Spark Open AI Model, Turning Mac Studio Pair into a Supercomputer ![Two Apple Mac Studio computers with keyboard and mouse on a wooden office desk](https://xab.info/media/2026/10/16/meta-otkryla-ii-muse-spark-open-dlya-mac/meta-otkryla-ii-muse-spark-open-dlya-mac-1.webp) ## Key Points - Meta has opened access to the weights of the multimodal Muse Spark Open model (816B total parameters, 42B active, 1M context window). - The model is optimized via the MLX framework for local execution on Macs with M5 Ultra chips, requiring at least 1 TB of unified memory. - The optimal configuration consists of two Mac Studio M5 Ultra units working as a single supercomputer with a throughput of 1.2 TB/s. - The model features optimized kernels for NVIDIA and AMD and is available on Hugging Face, Google Cloud Model Garden, Microsoft Foundry, and OpenRouter. - In tests, the AI agent autonomously edited the Mastodon Ice Cubes client code and performed vertical editing in DaVinci Resolve. Meta has opened public access to the weights of its newest multimodal AI model, Muse Spark Open. A key feature of the release is the targeted optimization of the system for Apple desktop computers powered by M5 Ultra chips: for the first time in its class, the model can run locally on desktop machines rather than exclusively in the cloud. Access to the weights is already available on several major platforms, making the system accessible to developers and researchers worldwide. ### Architecture and Model Parameters Muse Spark Open is a multimodal model with a total of 816 billion parameters, of which 42 billion are actively engaged during each generation step. This "sparse" approach combines the expressiveness of a large model with reasonable computational resource requirements. The model supports a 1-million-token context window, enabling work with extremely long documents, codebases, and video streams without losing coherence. ### Local Execution on Apple Silicon Optimization is achieved through the Apple MLX framework, enabling execution on Apple Silicon chips. Full local operation of the system requires at least 1 terabyte of unified memory. As an optimal configuration, the developers recommend a setup of two Mac Studio computers powered by M5 Ultra processors: the unified memory (up to 512 GB per device) and a throughput of 1.2 TB/s allow the two workstations to function as a single supercomputer. ### NVIDIA and AMD Support and Distribution Channels In addition to local execution on Apple Silicon chips, the model features specially optimized compute kernels to accelerate performance on NVIDIA and AMD graphics processors. This expands the range of hardware platforms where Muse Spark Open can be deployed. At launch, the model is available on Hugging Face, Google Cloud Model Garden, Microsoft Foundry, and OpenRouter. ### Real-World Autonomy Tests During real-world scenario testing, Meta demonstrated the high autonomy of Muse Spark Open when paired with the Muse Code terminal. The AI agent completely and independently edited the code of the open-source iOS Mastodon client called Ice Cubes: fixing bugs, adding new features, and building a fully functional app. Furthermore, the model was utilized for autonomous tasks in the DaVinci Resolve video editing software—the AI analyzed the video footage frame by frame, highlighted the best clips, and performed automatic framing and editing for a vertical format. ### Why Meta is Releasing the Weights According to Meta representatives, releasing open-weight models allows developers to reduce deployment costs, ensures a high level of transparency, and provides the freedom to customize the system for specific needs using the PyTorch framework. Thus, the company is betting on the ecosystem surrounding the open model rather than just closed cloud services. ## Fact-Check & Sources - [Превращает настольные Mac в суперкомпьютер: Meta открыла ИИ Muse Spark Open](https://www.rbc.ua/ukr/news/peretvoryue-nastilni-mac-superkomp-yuter-1792136432.html) - Primary source: model parameters (816/42B, 1M tokens), MLX optimization, 1 TB RAM requirement, dual Mac Studio M5 Ultra setup (1.2 TB/s), NVIDIA/AMD kernels, distribution platforms, tests with Ice Cubes and DaVinci Resolve. ## FAQ ### How many parameters does the Muse Spark Open model have? It has a total of 816 billion parameters, with 42 billion actively engaged. The model is multimodal and supports a 1-million-token context window. ### What are the hardware requirements for local execution? Full local execution requires at least 1 terabyte of unified memory. The optimal configuration is two Mac Studio computers with M5 Ultra chips (up to 512 GB of memory per device) featuring a 1.2 TB/s throughput. ### Where can I access the model? Muse Spark Open is available on Hugging Face, Google Cloud Model Garden, Microsoft Foundry, and OpenRouter. In addition to Apple Silicon, the model includes optimized kernels for NVIDIA and AMD chips.