Meta releases Muse Wonder and Muse Wonder Full, calling them its first frontier models
Meta introduced Muse Wonder and Muse Wonder Full for agentic work, coding, and expert domains. Wonder is available via the Meta Model API and Muse Code; Wonder Full is limited to Muse Code subscribers.
Photo: LPS.1 / Wikimedia Commons, CC0. Meta logo via Wikimedia Commons (public domain).
Meta on October 12 introduced Muse Wonder and Muse Wonder Full, describing them as its most capable models to date for personal and professional agentic tasks, from navigating an insurance claim to building a working app. The company said the release builds on groundwork laid with Muse Spark and marks fourteen months since it began overhauling its AI efforts.
According to Meta, the models include Active Memory, which carries relevant user context across conversations, and Active Focus, which keeps working toward a goal when a task gets hard, tries other approaches, recovers from errors, and checks in before important actions. Meta said they bring capability to expert domains including law, finance and health, along with frontier coding performance.
Muse Wonder is available through the Meta Model API and Muse Code, while Muse Wonder Full is available for Muse Code subscribers. Meta said Muse Wonder will soon power experiences in Muse, rolling out first to subscribers, and pointed readers to its research blog for how the models were built and tested.
Meta said it trained the family to avoid sharing private information without permission, refuse help with actually harmful acts, and support defensive cybersecurity work, including letting developers use the models to find flaws in their own code. A fuller safety report, it said, will be available soon.