Meta Puts Muse Wonder’s Memory and Focus Skills Into Muse and Meta AI

The assistant in Meta’s Muse app is now 2.4 times less likely to forget critical information people need to reach their goals, Meta says, after the company moved Muse and Meta AI onto new models trained using Muse Wonder, its most capable model family.

Meta described the change Friday in a Meta AI Research blog post, calling the new models specialized versions of its existing ones. It did not name them or say how quickly the rollout will reach every account. Mark Zuckerberg put it more broadly in a post on X, writing that the upgrade brings “Active Memory and Active Focus from Muse Wonder to everyone.”

Two flagship features reach the everyday apps

Active Memory and Active Focus debuted with Muse Wonder and Muse Wonder Full on Oct. 12. At launch, Wonder was available through the Meta Model API and Muse Code, and Wonder Full was limited to Muse Code subscribers. That announcement also said Meta would later use Wonder “to develop specialized models that we can bring to everyone using Muse.” Friday’s release is that follow-through.

Active Memory is meant to carry useful details from one conversation to the next. Meta said it trained the models to tell temporary information from permanent information, to reason across time and to separate what is still relevant from what has gone stale. On Meta’s internal Active Memory evaluation, the current Muse model correctly used critical context in 90% of tasks, compared with 76% for the previous Muse model.

Active Focus goes after a different weakness, an agent that stalls when something breaks. Meta said it trained the models on “a variety of bottlenecks reflective of the real world” so they can try harder, recover or find another route. In Muse, the company reported a 67% improvement in completing everyday tasks where unexpected issues come up, without giving the baseline completion rate behind that number.

How Muse and Meta AI should behave differently

For Muse, Meta said the models learned when to push ahead on a user’s goal on their own and when to work through it live with the user, and they can take corrections in the middle of a running task without getting confused. In Meta AI, the company expects the same training to help the assistant break complex problems into clear steps, ask sharper follow-up questions and keep pace when someone changes direction mid-conversation. Meta published benchmark figures only for Muse, and all of them come from its own evaluations.

A longer memory also raises the question of what Muse keeps about people. Last month, WebProNews reported on Muse instructions that security researcher Karan Joshi extracted and shared with Wired, which showed the agent creating “a page for every person in the user’s life.” Friday’s post explains how the new models judge what is still relevant, but it does not mention any new controls for users to review or delete what the assistant remembers.