Meta launched its new open-weight AI model, Muse Glimmer, on August 10, 2026. The model is designed as a smaller agentic AI that can run efficiently on a Mac or PC equipped with a single graphics card, making it accessible for local deployment and customization by users [1, 2]. Open-weight models like Muse Glimmer have publicly accessible core components, enabling developers to adjust and run the model without centralized control [1, 2].
This release follows the formation of Meta’s superintelligence team in 2025, created to help Meta regain its edge in AI research and development [1]. Meta also announced plans to soon release an open-weight version of a more powerful model called Muse Spark [2].
Alongside the Muse Glimmer launch, CEO Mark Zuckerberg published a 6,500-word essay advocating for open-weight AI development. He warned against excessive US regulatory restrictions, which he said disadvantage American labs compared to foreign competitors. Zuckerberg wrote, "Foreign labs currently hold several advantages here since American labs have to comply with many additional restrictions on training data. US policy must reduce this additional friction if we want American open-source models to lead over time" [1]. He argued that successful US AI leadership would increase prosperity and security for American communities compared to geopolitical rivals [3].
Zuckerberg’s call for policy reform highlights a competitive environment where Chinese startups like Moonshot, Alibaba, and DeepSeek currently lead in open-weight AI performance, challenging US labs [1].
Meta’s stock, which had fallen about 10% earlier in 2026, rose approximately 1% in pre-market trading on August 10 after the Muse Glimmer announcement [1].
Meta’s next step is the planned release of an open-weight version of the stronger Muse Spark model, which will further expand its open AI offerings [2].
