Meta launches Muse Wonder as a frontier model for agents

Meta is introducing Muse Wonder and Muse Wonder Full, a new model family the company describes as its most capable yet for personal and professional agentic tasks. The company says the models are built for work such as coding, understanding leases, handling insurance claims and reasoning through financial or health-related information.

The practical shift is that Meta is putting more emphasis on agents that keep context and pursue longer tasks. It describes two key capabilities: Active Memory, which carries relevant details across conversations, and Active Focus, which helps the model recover from errors, try alternate approaches and ask for clarification before important actions.

Access is split by product. Muse Wonder is available through the Meta Model API and Muse Code, while Muse Wonder Full is available to Muse Code subscribers. Meta also says Muse Wonder will power more of the Muse agent experience, first for subscribers.

Independent coverage frames this as Meta’s first serious frontier-model push after rebuilding its AI organization around Meta Superintelligence Labs. TechCrunch reports that Meta positions the Wonder models as larger and more capable than Muse Spark, with stronger coding and complex-task performance. The caution is that Meta’s frontier claim still needs real-world testing outside company benchmarks.