Updated

Muse Wonder and Muse Wonder Full

Meta introduced Muse Wonder and Muse Wonder Full, multimodal models for personal and professional agentic tasks that use Active Memory and Active Focus. Muse Wonder is available through the Meta Model API and Muse Code; Muse Wonder Full is available to Muse Code subscribers.

By No Code MBA · Facts checked

Provider
Meta
Announcement date
Status at last check
Generally available
Model family
Muse
Preceding model
Muse Spark 1.3
Capabilities
Text, Coding

What changed

Muse Wonder and Muse Wonder Full are Meta's largest Muse models so far, trained at its Prometheus data center on larger base models than earlier releases. They add Active Memory, which keeps user context across conversations, and Active Focus, which keeps working on hard goals by trying other approaches and checking in before important actions. Meta reports stronger results on agentic, coding, legal, finance and health tasks, and better generalization to unfamiliar tasks than the Muse Spark 1.3 trend predicts.

Access and availability

Muse Wonder is available through the Meta Model API and Muse Code, and is rolling out in the Muse app to subscribers first. Muse Wonder Full is available only to Muse Code subscribers.

Official access documentation

What you can build

Compare the models on a small set of multi-step workflows that require recalling earlier context, using tools, and recovering from a blocker. Track task completion and review consequential actions before relying on them.

Limitations

Meta did not state context window, supported modalities or pricing in the announcement, and its full safety report was listed as coming soon. Muse Wonder Full is limited to Muse Code subscribers.

Sources and release history