Meta is accelerating the rollout of its Muse Spark model, a new foundation model built by Meta Superintelligence Labs designed to power a more responsive and context-aware AI experience across its ecosystem, including its Meta AI app and wider family of products.
First announced in April 2026, Muse Spark is described by Meta as its most powerful model yet, built to prioritise speed, multimodality and integration with the company’s social platforms.
The model is now underpinning faster voice interactions in the Meta AI app, alongside expanded features that allow users to generate images, access recommendations and receive real-time contextual assistance based on what they see through their devices.
The latest update brings what Meta calls “live AI” capabilities into broader use, allowing users to point their camera at the world and ask questions about their surroundings in real time. The feature builds on existing functionality in Meta’s AI glasses and extends it into the mobile app, enabling more interactive, visually grounded conversations.
Muse Spark also introduces a more flexible voice experience, allowing users to interrupt conversations, switch topics and move between languages naturally during interactions.
Meta said the system can also surface content and recommendations from across its platforms, including Reels, maps and other contextual signals, to deliver more personalised responses.
Shopping functionality is also being expanded, with Meta AI now able to surface listings from Facebook Marketplace alongside broader internet results. Users can refine searches by price, style or proximity, and view items on a map to better understand local availability.
According to Meta, Muse Spark will progressively roll out across WhatsApp, Instagram, Facebook, Messenger and Threads in the coming weeks, as well as to its AI glasses range.
The company is also testing new integrations such as “side chats” in group conversations and @meta.ai mentions, designed to bring AI assistance directly into social interactions.
Meta said the goal of Muse Spark is to create a more “personal superintelligence” that understands users in context, combining reasoning, visual perception and social signals to make interactions more useful and immediate across its apps and devices.
