Menlo Park: Meta has launched a new artificial intelligence model as the company makes a strong return to its earlier approach of giving developers wider access to AI technology. The new model, called Muse Glimmer, is designed to perform smaller tasks independently and can run on personal computers using a single graphics card. The move is part of a broader effort by Meta to make powerful artificial intelligence more accessible outside large technology companies and their data centres.
The launch comes as Meta CEO Mark Zuckerberg argues that the future of artificial intelligence should not be controlled by a small number of companies. In a lengthy statement about the future of AI, Zuckerberg said advanced technology should be made available to a much wider group of people. He argued that giving developers access to AI models can encourage innovation, reduce costs and allow users to build systems suited to their own needs.
Muse Glimmer has 30 billion parameters, making it considerably smaller than some of the most powerful AI systems being developed by leading technology companies. However, its smaller size is also one of its main advantages. The model has been developed with tasks that can be carried out directly on personal devices in mind.
This could make AI more useful for people and businesses that do not want to depend completely on cloud based services. Running an AI model locally can provide greater control over information and may also reduce the cost and delay involved in sending requests to distant data centres.
The model is particularly focused on what is known as agentic AI. These systems are designed not only to answer questions but also to complete a series of tasks with less direct human involvement. They can be used for activities such as planning, coding, research and other computer based work.
Meta is also preparing a more significant step. The company plans to release the weights of Muse Spark 1.2, which it describes as its most advanced AI model so far. The release could attract considerable attention because it would give developers access to a much more capable system than Muse Glimmer.
The decision is important because Meta had moved toward a more controlled approach with some of its recent AI development. The return to open weight models suggests the company now sees wider access as an important part of its strategy.
The announcement also has a strong international dimension. Zuckerberg has argued that the United States needs to avoid putting excessive restrictions on open AI development if it wants to maintain its advantage over China. Chinese companies have made rapid progress in artificial intelligence and have released models that developers around the world can use and modify.
Meta believes American developers could lose ground if they face heavier restrictions while competing systems from China remain widely available. Zuckerberg has therefore called for policies that allow American companies and developers to build and deploy open AI systems more easily.
The debate is not only about technology. It is also about who should control the future of artificial intelligence. Companies such as OpenAI, Google and Anthropic have invested heavily in highly controlled AI systems that are mainly accessed through their own products and services. Meta is presenting a different model in which developers can obtain the weights, study them, modify them and build their own applications.
However, open AI also brings concerns. Making powerful models widely available can create questions about security, misuse and accountability. Critics argue that releasing increasingly capable models without sufficient safeguards could make it harder to control harmful applications.
Meta is also trying to address concerns surrounding the rapid expansion of its AI infrastructure. Zuckerberg announced a one billion dollar fund for communities affected by the company's growing data centre operations. Meta is investing heavily in computing infrastructure to support its AI ambitions, but large data centres can create concerns about electricity use, water consumption and their wider impact on local communities.
The latest move therefore represents more than the launch of another AI model. Meta is trying to position itself as a major supporter of open AI while competing with some of the world's most powerful artificial intelligence companies.
The success of this strategy will depend heavily on the performance of its new models. If Muse Spark 1.2 proves capable of competing with leading closed AI systems while remaining openly available, Meta could give developers a powerful alternative to the increasingly expensive and controlled AI services offered by its rivals.
For now, Muse Glimmer is the first major sign that Meta is once again placing open AI at the centre of its strategy. The company is betting that wider access, local AI and a large community of developers can help it strengthen its position in the rapidly changing global artificial intelligence race.