Meta introduced Muse Wonder and Muse Wonder Full on October 12, expanding its AI lineup with models aimed at personal and professional agent tasks. Muse Wonder is available through the Meta Model API and Muse Code, while Muse Wonder Full is accessible to Muse Code subscribers. Integration into the Muse application is planned, beginning with subscribers.
The launch places greater emphasis on completing work across multiple steps. Meta describes applications ranging from interpreting documents to building software, with two features supporting longer tasks. Active Memory is designed to retrieve relevant information across conversations, while Active Focus is intended to help models persist, recover from errors and seek clarification when necessary. These descriptions represent Meta’s stated capabilities rather than independent findings.
Meta also says it trained the models to protect private information and refuse harmful assistance. The company indicated that further safety evaluations would be published in a forthcoming report.
From an editorial perspective, the release makes task completion a useful measure of progress. A successful agent must do more than produce a persuasive answer: it must retain the right context, recognize failed steps and know when to request human input. These qualities will matter when developers assess whether a model can support dependable workflows.
The immediate development is the availability of two new models through specified channels. Their broader reliability, comparative performance and suitability for particular business processes will require evidence from practical deployment.
Source: https://about.fb.com/news/2026/10/introducing-muse-wonder-models/