About

Meta Launches Muse Wonder Models, Claims Top Agent Benchmarks as Muse Downloads Slow

Published
Score
30

Why it matters

Meta announced the Muse Wonder series on October 12, 2026. Muse Wonder is available through its developer API and Muse Code. Muse Wonder Full is limited to Muse Code subscribers. Meta says both models lead on personal and professional agentic tasks, including handling insurance claims, reading leases, and building apps. The release adds "Active Memory," which is meant to retain context across conversations, and "Active Focus," which is meant to keep the model working toward longer goals. Meta says the models were trained at its Prometheus data center and will later power the consumer Muse app, subscribers first. Meta Superintelligence Labs, which builds the Muse family, leads the effort. Mark Zuckerberg calls Muse "the centerpiece of our vision."

The announcement included no numerical benchmark scores, so independent comparison with OpenAI, Anthropic, and Google models is not currently possible. The Muse personal agent app launched September 8. It hit No. 1 on the U.S. App Store and passed five million downloads in 22 days, according to Sensor Tower estimates. Since then, average daily downloads have fallen about 8.1% week over week to roughly 165,000, and net daily user growth has slowed about 62% from late September. TIME reported that Muse builds continuously updated profiles of users, and Meta did not dispute the finding. Hunterbrook Capital, a short seller that took a position against Meta on October 9, has also warned that the agent could be prompted to compile lists of accounts tied to vulnerable groups. No regulatory action specific to this release has been reported. One report cites mounting regulatory scrutiny of Meta's AI push more broadly.

Meta is using a model upgrade to address a visible slowdown in adoption, and retention reportedly trails ChatGPT. Analysts say better benchmarks alone will not fix either the slowdown or the permissions concerns. Attorneys tracking this should watch the profiling and data-permissions issues, since they carry the clearest regulatory exposure. Indicators to watch include independent evaluations, daily active users, paid conversion, and Meta's upcoming quarterly results.

Sources

mail Subscribe to Artificial Intelligence email updates

Primary sources. No fluff. Straight to your inbox.

Also on LawSnap