Meta is taking a new step in its AI strategy with the introduction of Muse Spark, the first model in the Muse family from Meta Superintelligence Labs.
With this model, the company aims to lay the foundation for what it describes as personal superintelligence—a form of artificial intelligence that not only answers questions but also actively thinks alongside users, understands them, and supports them in their daily lives.
Muse Spark comes at a crucial moment for Meta. Reuters notes that this is the first AI model the company has unveiled in about a year, following earlier unmet expectations surrounding the Llama 4 models. As such, the launch represents not only a technological advancement but also an effort to reconnect with the leading pack in the AI market.
The model stands out because it was designed from the ground up as a multimodal reasoning model. Muse Spark can combine text, images, and other data types and also supports the use of external tools and the deployment of multiple AI agents simultaneously. Meta is thus positioning the system as the next step toward AI that can independently perform complex tasks and better understand context.
The introduction of Muse Spark also marks a broader overhaul of Meta’s AI approach. According to the company, work has been underway in recent months on a new technical stack designed to scale more efficiently, ranging from model architecture and training to infrastructure such as the Hyperion data center. Meta claims that it can achieve comparable performance with significantly less computing power than previous models.
Meta is heavily investing in its superintelligence team
Reuters adds that this technological leap involves substantial investments. Last year, Meta assembled a new superintelligence team to catch up with competitors, including through a multi-billion-dollar investment in Scale AI and the appointment of CEO Alex Wang. According to the news agency, some engineers were lured with exceptionally high compensation packages, underscoring the company’s strategic priority for AI.
In terms of performance, Meta claims that Muse Spark is competitive across visual analysis, reasoning, and healthcare applications. At the same time, independent evaluations show that the model is not yet at the top in every area. According to external benchmarks, Muse Spark performs strongly in language and visual understanding, but lags behind competitors in programming and complex abstract reasoning. On a broad ranking by the evaluation platform Artificial Analysis, the model is tied for fourth place.
An important new feature is the Contemplating mode, which allows multiple AI agents to work on a problem in parallel. This approach is intended to enable deeper analysis without significantly increasing response time. Meta sees this as a way to compete with advanced reasoning modes of other leading AI systems.
Meta is focusing on practical AI applications
The applications Meta has in mind go beyond traditional chat interfaces. Muse Spark can analyze images, recognize objects, and assist users with practical tasks. The model also plays a role in the field of health, for example, by extracting nutritional information from images or explaining bodily processes. To this end, Meta has collaborated with numerous doctors to improve the quality of its answers.
At the same time, the launch offers more insight into how Meta plans to monetize AI. The company hints at e-commerce integrations, allowing users to view and purchase products directly in the chatbot. Additionally, Meta is strongly focused on increasing engagement within its existing ecosystem, which now includes billions of users.
The rollout of Muse Spark is taking place in phases. Initially, the model will be available via the Meta AI app and website, after which it will replace the existing Llama models within platforms such as WhatsApp, Instagram, and Facebook in the coming weeks. Notably, Meta is not sharing details about the model’s scale and is deviating from its previous open strategy by making only a limited API preview available to selected partners.