Meta challenges rivals with new closed Muse Spark AI model

Meta has introduced a new large language model called Muse Spark AI , positioning it as a proprietary system intended to run the company’s internal chatbot and other integrated product features. The company announced the model on Wednesday, describing it as a closed-source release that differs from Meta’s earlier approach of making model code and architecture publicly available.

The launch is Meta’s first major AI model release in more than a year, and it signals a strategic pivot toward tighter control over distribution and usage. Instead of broad public access, Meta said it will offer a limited private preview to select partners through an application programming interface (API), enabling developers to incorporate Muse Spark into their own software under controlled terms.

Meta’s decision arrives as competition among leading AI developers remains intense. The company is seeking to narrow the gap with major AI labs including OpenAI , Anthropic , and Google DeepMind , which have continued to release new models at a fast pace and have built a strong position in generative AI.

Meta has invested billions of dollars in recruiting AI talent and expanding infrastructure, and the company is using this release to demonstrate progress in a market where product cycles and developer adoption can move quickly.

Company officials indicated that some versions of Muse Spark could be open-sourced at a later date, but the initial rollout prioritizes internal deployment and restricted partner access. By keeping the model closed at launch, Meta retains greater oversight of how the system is used, how it is integrated into products, and how it could be monetized in the future, according to the company’s framing of the strategy.

Meta CEO Mark Zuckerberg linked the model to a broader long-term ambition in a post on the Threads platform. ” Zuckerberg said the concept is focused on helping individuals achieve personal goals, and he tied that direction to Meta’s plan to embed advanced AI capabilities across its app ecosystem, including Instagram , Facebook , and WhatsApp .

How Muse Spark performs and how widely it is adopted will be watched as a measure of whether Meta can translate its large investments into products that compete at the top tier of the market. Key uncertainties remain around the scope of partner access during the private preview, the timeline for any potential open-sourcing of versions of the model, and how quickly Meta can scale features powered by Muse Spark across its platforms while maintaining the controlled approach it has outlined.