New model from Meta strengthens multimodal intelligence, faster reasoning, and ecosystem integration across its platforms.
Meta Superintelligence Labs has introduced Muse Spark, its first large language model in a new generation of AI systems designed to bring more personalised, context-aware intelligence to users across its ecosystem. The launch marks a significant step in Meta’s ambition to build what it describes as “personal superintelligence” — an AI assistant capable of understanding and supporting users in real-world scenarios.
Muse Spark is engineered specifically for Meta’s family of applications, including its AI assistant, and is expected to enhance performance across platforms such as WhatsApp, Instagram, Facebook, Messenger, and emerging AI-powered devices. The model is already powering the Meta AI app and web experience, with broader rollout planned in the coming weeks.
At its core, Muse Spark represents a shift in how Meta approaches AI development. Built over nine months during which the company rebuilt its AI stack, the model is part of a structured “Muse” series designed to scale progressively. While this first iteration is intentionally lightweight and fast, it is capable of handling complex reasoning tasks across domains such as science, mathematics, and health.
A key advancement lies in its multimodal capabilities. Muse Spark enables Meta AI to interpret and analyse images alongside text, allowing users to interact with AI in more intuitive ways. For example, users can capture images of products or environments and receive contextual insights without needing detailed text input. This approach reflects a broader move toward AI systems that can “see” and understand the world, rather than relying solely on written prompts.
The model also introduces parallel task processing through multiple subagents. This allows Meta AI to handle complex queries more efficiently by breaking them into simultaneous workflows — such as planning travel, comparing destinations, and recommending activities — delivering faster and more comprehensive responses.
Healthcare is another area of focus. Recognising that health-related queries are among the most common uses of AI, Meta has collaborated with medical professionals to enhance Muse Spark’s ability to provide reliable and informative responses, including interpretation of charts and images where relevant.
In addition to reasoning and perception, Muse Spark expands creative and functional use cases. The model supports visual coding, enabling users to generate websites, dashboards, and interactive experiences directly from prompts. It also integrates with Meta’s social ecosystem to provide personalised recommendations in areas such as shopping, lifestyle, and local discovery, drawing on content shared across its platforms.
The introduction of “shopping mode” further demonstrates Meta’s strategy to blend AI with its existing creator and community-driven ecosystem.
By surfacing recommendations based on user interests and social signals, the feature aims to deliver a more personalised discovery experience. While currently rolling out in the US, expansion to other markets is expected.
Looking ahead, Meta plans to extend Muse Spark’s capabilities through API access for select partners and potentially open-source future iterations. The company also emphasises continued investment in safety, privacy, and risk frameworks as the technology evolves.
Muse Spark is positioned as an early milestone in a broader roadmap. With larger and more advanced models already in development, Meta is signalling its intent to compete aggressively in the next phase of AI innovation — one defined not just by intelligence, but by relevance, context, and integration into everyday life.
The launch underscores a growing industry shift toward AI systems that are deeply embedded in user ecosystems, capable of understanding behaviour, preferences, and environments. For Meta, Muse Spark is not just a model, but the foundation of a more connected and personalised AI future.
