Meta Launches Muse Glimmer AI Model as Zuckerberg Pushes for Open-Weight AI in US

Meta has introduced Muse Glimmer, a compact open-weight AI model, as Mark Zuckerberg urges fewer US restrictions to help American developers compete with Chinese AI companies.

Meta Launches Muse Glimmer AI Model as Zuckerberg Pushes for Open-Weight AI in US

Meta Introduces New Open-Weight AI Model

Meta has unveiled Muse Glimmer, a new open-weight artificial intelligence model designed to perform agentic tasks while operating on relatively modest computing hardware.

The launch comes as Meta seeks to strengthen its position in the increasingly competitive global AI market. The company also indicated that additional open-weight models could be introduced in the future.

Unlike some of the largest AI systems developed by leading technology companies, Muse Glimmer is designed to be considerably smaller and can reportedly run on a Mac or PC equipped with a single graphics card.

Muse Glimmer Targets On-Device AI

A major focus of the new model is enabling AI capabilities to operate directly on users' devices. Running AI locally can offer potential advantages, including reduced dependence on cloud infrastructure and greater flexibility for users and developers.

Muse Glimmer is particularly aimed at agentic AI tasks, where systems can perform multiple steps or actions based on user instructions.

The smaller size of the model could make it more accessible to developers and organisations that do not have access to large-scale computing infrastructure.

Zuckerberg Calls for Changes to US AI Policy

Meta CEO Mark Zuckerberg has renewed his support for open-weight AI models and called for a reassessment of US policies governing their development.

Zuckerberg argued that American AI companies face additional restrictions, particularly concerning training data, which could make it more difficult for them to compete with overseas developers.

He said reducing these barriers would be important if US-developed open-weight models are to remain competitive globally.

According to Zuckerberg, simply limiting access to foreign open-weight models would not provide an effective long-term solution.

Why Open-Weight AI Is Gaining Attention

Open-weight AI models generally provide developers with access to important components that allow models to be downloaded, customised and adapted for different applications.

They can also be less expensive to operate than some proprietary frontier models.

This flexibility has attracted businesses that are looking for ways to control rising AI costs while adapting models to specific requirements.

By comparison, closed AI systems typically keep their underlying model components under the control of the developer, limiting how users can customise or deploy them.

Chinese AI Companies Increase Competition

The debate comes as Chinese AI companies make significant progress in open-weight models. Models from companies such as Moonshot, Alibaba and DeepSeek are increasingly competing with systems developed by major US AI companies.

Chinese models including Kimi K3, Qwen3.8-Max and DeepSeek V4-Flash have emerged as notable competitors in the open-weight segment.

Meanwhile, leading AI developers in the US, including OpenAI, Anthropic and Google's parent company Alphabet, have largely focused on closed models.

This growing competition has increased pressure on US technology companies to maintain leadership in AI development.

Meta Strengthens Its AI Strategy

Meta's latest model launch comes after the company invested heavily in building its artificial intelligence capabilities. The social media company created a dedicated superintelligence team as part of a broader effort to compete more aggressively in advanced AI.

Meta has been pursuing different approaches to AI development, including open-weight models, while also investing heavily in computing infrastructure and talent.

The company's latest release reflects its strategy of making smaller and more accessible AI systems available to developers.

Businesses Look for Lower AI Costs

Open-weight models are gaining traction as companies become increasingly concerned about the cost of deploying advanced AI systems. Businesses using AI at scale can face significant expenses related to computing resources, model access and cloud infrastructure.

Smaller models that can be deployed locally could provide an alternative for certain applications.

Open models can also give companies greater control over how AI systems are customised and integrated into their existing technology infrastructure.

Cybersecurity Concerns Add to the Debate

The discussion around open and closed AI models has also been influenced by cybersecurity concerns.Recent incidents involving AI systems have highlighted both the potential benefits and risks associated with increasingly capable models.

Some organisations argue that open models can offer greater flexibility for cybersecurity research because developers have fewer restrictions on how the technology can be used.

The debate has therefore expanded beyond cost and performance to include questions surrounding security, access and responsible AI development.

Zuckerberg Supports AI Model Distillation

Zuckerberg has also expressed support for AI model distillation, a technique in which a smaller model learns from the capabilities of a more powerful system.

This approach can help create more efficient AI models while potentially reducing the computing resources needed for deployment.

Smaller models could be particularly useful for applications where running large AI systems through cloud infrastructure is impractical.

Meta Plans Stronger Governance for AI Releases

Alongside its push for open-weight models, Meta plans to strengthen oversight of its AI releases. Zuckerberg said the company would establish a governance structure that gives independent directors authority to approve safety criteria before models are released.

The move is intended to create additional oversight around the risks associated with increasingly capable AI systems.

US-China AI Competition Intensifies

Meta's latest move highlights the growing competition between the United States and China in artificial intelligence. While US companies continue to lead in several areas of advanced AI development, Chinese companies are making rapid progress, particularly in open-weight models.

The policy debate in Washington is therefore increasingly focused on how to encourage innovation while maintaining appropriate safeguards.

Conclusion

Meta's launch of Muse Glimmer highlights the growing importance of smaller, flexible and open-weight AI models. Mark Zuckerberg's call for fewer policy barriers reflects Meta's broader push to strengthen US competitiveness as Chinese AI developers rapidly advance in the open-model space.