☰
✕
UA RU EN

Meta Launches Muse Spark Open: An 800 Billion Parameter AI Model Tailored for Mac Studio with M5 Ultra

Meta представила новий потужний AI-модель Muse Spark Open для Mac Studio з 800 мільярдами параметрів. Photo: НВ — Техно

Introducing Muse Spark Open by Meta

Meta has unveiled Muse Spark Open, an advanced AI model boasting 800 billion parameters, specifically optimized for Apple’s Mac Studio equipped with the M5 Ultra chip. This cutting-edge model outperforms prominent Chinese AI systems such as GLM 5.3 and Kimi K3 in tasks involving AI agents and visual data analysis. Running Muse Spark Open locally demands a combined memory capacity of 1 TB, which requires multiple Mac Studio units working in tandem.

Each Mac Studio with an M5 Ultra processor supports up to 512 GB of RAM, making it necessary to link at least two machines to meet the 1 TB memory requirement. Alternatively, a setup of four Mac Studios with 256 GB each can be used. The estimated cost for such a configuration is around $38,000 (approximately 1.6 million UAH). Apple is expected to release the 512 GB Mac Studio model by the end of October.

Competitive Edge and Implications

Meta describes Muse Spark Open as the most powerful open-weight model developed in the United States. The model’s weights are crucial as they dictate how it processes information and generates responses. Muse Spark Open is available for download via the Hugging Face platform. Despite the steep hardware investment limiting accessibility for individual developers, large organizations might find purchasing multiple Mac Studios a viable alternative to cloud-based AI services. This development marks a significant milestone in advancing on-premises AI solutions.

The release of Muse Spark Open highlights the intensifying rivalry in AI development between Western and Eastern technology firms. While the high cost is a barrier, this new Meta model offers a valuable resource for enterprises aiming to integrate state-of-the-art AI capabilities into their workflows. Furthermore, it could drive increased investment in local AI infrastructure and foster growth in handling large-scale data processing.

As Meta continues to innovate in AI, the recent introduction of new Muse models featuring active memory and enhanced focus for complex tasks further emphasizes the company's commitment to advancing technology. These developments not only showcase Meta's competitive edge but also underline the growing demand for sophisticated AI solutions across various industries.