Meta brings Muse Wonder’s memory and task-persistence features to all Muse and Meta AI users
On 6 November Meta said it has trained specialized versions of its existing models with its flagship Muse Wonder and is rolling them out to everyone using Muse and Meta AI, bringing Active Memory and Active Focus. Meta says the Muse models forget critical information 2.4x less often and are 67% better at finishing everyday tasks that hit unexpected problems; all figures are Meta’s own.

In a research blog post, Meta says Muse Wonder, introduced a few weeks earlier, is its most capable model, and that it has now used it to train specialized versions of the existing models behind Muse and Meta AI and is rolling them out to people who use both products. Users get the new Wonder-trained models, not Muse Wonder itself.
Active Memory lets the assistants tell temporary from lasting information and recognize when earlier details become relevant again or go stale. Meta says the new Muse models show a 2.4x reduction in forgetting critical information, and on its Active Memory eval the share of tasks that correctly use critical context rose from 76% to 90%. Active Focus lets Muse keep trying, recover or find alternatives when a task runs into unexpected obstacles; Meta reports a 67% improvement on such everyday tasks. The post also lists its Personal Superintelligence Index rising from 38% to 47%.
For Meta AI the post gives no numbers, saying only that it breaks complex problems into clear steps, asks sharper follow-up questions and keeps up when users change direction. All metrics come from Meta’s internal evaluations; the post does not describe methods, sample sizes or rollout pace, and mentions no pricing change.
Sources
- Bringing Specialized Models to Muse and Meta AI — Meta AI Research