Meta introduced Muse Wonder and Muse Wonder Full on Oct. 12, positioning the models as its most capable systems yet for AI agents handling personal and professional tasks, while making the stronger version available only to Muse Code subscribers.
The company says the models can tackle multi-step work such as navigating an insurance claim, interpreting a lease, handling legal or financial information, and building software. Meta says a feature called Active Memory can carry useful context across conversations and retrieve it when relevant to a task. Another, Active Focus, is designed to keep working through setbacks, try alternative approaches, and ask for clarification or confirmation before important actions.
Muse Wonder is available through Meta’s Model API and Muse Code; Muse Wonder Full is currently limited to Muse Code subscribers. Meta says the standard model will later power features in its Muse personal-agent app, with an initial rollout to subscribers. The launch also intensifies Meta’s competition with AI developers such as OpenAI and Anthropic, but the company’s performance comparisons are based on its own tests, according to TechCrunch.
CNBC reported that access to Wonder through the Muse app requires a paid subscription, as Meta seeks revenue from AI beyond advertising. Meta says it built safeguards into the models to protect private information and reduce harmful use, while questions about how personal agents handle users’ messages and other data remain part of the rollout.