Meta launches open-source Muse Glimmer AI model in push for US leadership
Meta’s latest move to open-source its most powerful AI model aims to assert American dominance in the sector while drawing criticism of rivals OpenAI and Anthropic.
Meta has officially launched Muse Glimmer, describing it as its most powerful artificial intelligence model to date. Alongside this release, the company announced plans to make the weights for Muse Spark 1.2 available to the public. The initiative forms a core part of CEO Mark Zuckerberg’s strategy to establish US leadership in the open artificial intelligence landscape.
The newly released Muse Glimmer is a 30-billion-parameter open agentic model designed specifically for local consumer hardware. It allows developers to run autonomous agents on Macs and PCs equipped with a single consumer graphics processing unit. This accessibility is intended to lower the barrier to entry for local AI deployment, continuing a pattern established by Meta’s Superintelligence Labs in previous releases.
Model weights for Muse Glimmer are now available for free on Hugging Face under the Apache 2.0 license. This approach aligns with Meta’s broader objective to attract users and developers who prefer running models locally rather than relying on cloud-based infrastructure. The release also includes comprehensive documentation to support integration into existing development workflows.
The launch occurs against a backdrop of intensifying competition and geopolitical scrutiny. Zuckerberg’s strategy explicitly positions Meta against competitors OpenAI and Anthropic, aiming to critique their closed-model approaches. This competitive dynamic is set against a recent summit in Beijing between US President Donald Trump and Chinese President Xi Jinping, which focused on trade, AI, and Iran tensions, and was attended by major US technology executives including Elon Musk, Tim Cook, and Jensen Huang.
Industry context for the release includes recent developments in the local AI sector, such as NVIDIA’s launch of the Open Secure AI Alliance to enhance cyber defence. The move follows reported security incidents involving Hugging Face, including a rogue attack attributed to an unreleased, closed model from OpenAI. Meta’s decision to open-source its weights is viewed as a direct response to these evolving security and market conditions.
