Meta introduced its latest family of artificial intelligence models on Monday, a collection of open-source systems known as Muse Glimmer that are designed to run on a laptop. The launch was accompanied by a 6,500-word essay from CEO Mark Zuckerberg laying out his case for American open-source AI.

In the essay, Zuckerberg contends that powerful AI should not be controlled by a small number of companies—an apparent critique of competitors such as OpenAI and Anthropic. He frames Meta's open-source commitment as essential to countering Chinese open-source models from DeepSeek and Moonshot, which he says have been approaching the American frontier. DeepSeek in particular drew wide attention earlier this year after releasing models that delivered near-frontier performance at a fraction of the training cost associated with leading Western systems, intensifying the competitive pressure on U.S. labs. Zuckerberg also called on Washington to back American AI efforts.

As part of the renewed open-source push, Meta said it plans to open the weights for Muse Spark 1.2—its most advanced AI model currently on the market. Opening the weights means Meta will publicly release the model's underlying trained parameters, allowing anyone to download, inspect, and modify them rather than being limited to access through Meta's own products or API. The open-weight approach has been a subject of ongoing debate within the AI community: proponents argue it accelerates innovation and enables independent safety scrutiny, while critics caution that publicly releasing powerful model weights could lower barriers to misuse.

"I do not believe restricting access to foreign open source models is an effective solution. Our goal should be for American open source models to be the best globally. This requires removing the hurdles that make it harder for American open source models to compete," Zuckerberg wrote.

The essay is part of a coordinated media blitz by Meta that included a Zuckerberg-authored opinion piece on the same themes published in the Wall Street Journal last month, along with interviews with select media outlets.

The Muse Glimmer release arrives in a U.S. market increasingly dominated by closed-source AI companies—firms that sell controlled access to their most powerful models. OpenAI, Anthropic, and Google generally keep their most advanced "frontier" models behind closed doors, maintaining tight control over both the weights and the infrastructure.

Meta originally sought to differentiate itself by embracing open source, launching the open-weight Llama family of models in February 2023. The Llama family expanded through several versions, making high-performing models available for outside developers to download and modify. This gave Meta influence over the broader AI ecosystem, even as rivals captured more revenue by selling access to proprietary models.

Over the past year, however, Meta's commitment to open-weight releases has blurred, with the company mixing open releases alongside proprietary ones. Meta has invested heavily in its new Superintelligence Labs, including bringing in former Scale AI CEO Alexander Wang. The company has been working to position the Muse family as the foundation for a more capable Meta AI assistant across Facebook, Instagram, WhatsApp, and its Ray-Ban Meta AI glasses, while also aiming to remain competitive with models from Anthropic and OpenAI.

The refocus on open-source may reflect an attempt by Meta to regain momentum after recent models fell short of systems from OpenAI and Anthropic. Meta is also looking to reassure investors that it remains competitive with frontier labs.

Meta said it trained the new Glimmer series using distillation, a process in which a smaller model learns from Muse Spark, Meta's larger proprietary system. Distillation has become a widely used technique for producing efficient models that retain much of the capability of larger systems while requiring far less compute to run. In his essay, Zuckerberg wrote that American AI developers face disadvantages compared with Chinese rivals, including in training data, and argued that policy should reduce those barriers.

With the strategy shift, Meta will likely be hoping that its open-source commitment will distinguish it from rivals who keep their most capable systems behind closed doors, while also giving the company a political argument for user and government support.

This story was originally featured on Fortune.com.