Meta launches Muse Wonder and Muse Wonder Full, its most capable models yet
Meta has introduced the Muse Wonder series, made up of Muse Wonder and Muse Wonder Full, which the company calls its most capable models to date. According to Meta, they deliver state-of-the-art performance on personal and professional agentic tasks.
- Meta introduced Muse Wonder and Muse Wonder Full, which it calls its most capable models to date.
- The models build on Muse Spark and target personal and professional agentic tasks, including law, finance, health and coding.
- New features include Active Memory, which carries context across conversations, and Active Focus, which keeps working through hard tasks.
- Muse Wonder is available through the Meta Model API and Muse Code, and will soon power more of Muse, first for subscribers.
- A full report with additional safety evaluations is promised soon.
Built on Muse Spark, aimed at real-world tasks
Meta says the Muse Wonder series builds on the groundwork laid with Muse Spark. The company describes the release as the result of an overhaul of its AI efforts that began fourteen months ago, moving from rebuilding its foundations to pushing the frontier.
The models are aimed at everyday problems where an AI has to interpret information and work through it. Meta gives examples such as navigating an insurance claim, understanding a lease, or making sense of someone's finances. According to the company, Muse Wonder and Muse Wonder Full bring that capability to expert domains including law, finance and health. Meta also points to frontier coding performance for people building anything from a game to an app for a new business.
Active Memory and Active Focus
Meta highlights two capabilities built into the new models, both designed for work that unfolds over time.
- Active Memory: the ability to carry what matters to the user across conversations and bring back details relevant to the task at hand.
- Active Focus: the ability to keep working toward a goal when a task gets hard, try other approaches, recover from errors and follow through. According to Meta, the model asks for clarification when needed and checks in before following through on anything important.
Safety approach
Meta notes that a personal agent can take on real tasks, which often means working across a user's messages, calendars and documents. For that reason, the company says it trained the Muse Wonder family to avoid sharing private information without permission, to turn down requests to assist in performing actually harmful acts, and to work to improve security and strengthen defenses against cyber attacks.
Meta adds that developers and security teams can use models like Muse Wonder to find flaws in their own code and strengthen their systems, and that it made sure the models can still do that work. The company says it trains, evaluates and releases its most capable models in line with the standards set out in its Meta Superintelligence Scaling Framework. A full report with additional safety evaluations will be available soon, according to Meta.
Availability and what comes next
Muse Wonder is available today through the Meta Model API and Muse Code. Meta says it will soon power more of Muse, rolling out first to subscribers. The announcement does not give details on pricing or on the availability of Muse Wonder Full beyond the series launch.
Meta describes the release as a major step toward personal superintelligence for everyone, with much more to come. For more detail on how the models were built, tested and what they can do, the company points readers to its research blog.
