Meta has introduced Muse Spark, a new large language model developed by its Meta Superintelligence Labs, marking a significant step toward building more advanced, personalized AI experiences across its ecosystem.
The model is the first in Meta’s new “Muse” series and has been designed to deliver faster performance while handling complex reasoning tasks across areas such as science, mathematics, and health. Despite being smaller and more efficient by design, Muse Spark represents the company’s most powerful AI model to date and serves as the foundation for future, more advanced iterations already in development.
Currently powering the Meta AI assistant on the Meta AI app and website, Muse Spark is set to expand across major platforms including WhatsApp, Instagram, Facebook, and Messenger, as well as emerging hardware like AI-enabled glasses. Meta also plans to provide early API access to selected partners through a private preview.
A key upgrade introduced alongside Muse Spark is a more capable Meta AI assistant, which now supports both quick responses and deeper problem-solving. The system can deploy multiple “sub-agents” simultaneously to handle different aspects of a task, improving speed and quality of responses for more complex queries.
The model also brings enhanced multimodal capabilities, allowing Meta AI to interpret images and visual inputs in addition to text. This enables users to interact with the assistant in more intuitive ways, such as analyzing products, comparing items, or extracting insights from real-world visuals. The feature is expected to play a significant role in areas like health, where users can receive more detailed and contextual guidance.
In addition, Muse Spark introduces new creative and functional capabilities, including visual coding that allows users to generate websites, dashboards, and mini-games directly from prompts. The assistant is also being integrated with social and shopping experiences, leveraging content and recommendations from across Meta’s platforms to provide more personalized suggestions.
Meta says the rollout of these features will begin in the United States, with plans to expand globally in the coming weeks. As the company continues to evolve its AI systems, it aims to integrate richer, more contextual content—such as posts, images, and videos—directly into user interactions.
The launch of Muse Spark underscores Meta’s broader ambition to develop “personal superintelligence”—AI systems that not only respond to user queries but deeply understand individual contexts, preferences, and real-world environments.
