Meta ships Muse Wonder-trained models to all Muse, Meta AI users
TL;DR
- Meta shipped specialized models trained with its Muse Wonder frontier model to all Muse and Meta AI users starting Friday.
- Meta claims the new Muse models are 2.4x less likely to forget critical information and 67% better at self-correcting everyday tasks.
- On Meta's internal evaluation, the new Muse models correctly use critical context in 90% of tasks, up from 76%.
Meta says the specialized models now reaching Muse are 2.4 times less likely to forget critical information and 67% better at self-correcting and completing everyday tasks. The New Stack reports that starting Friday, everyone using the Muse agent and Meta AI assistant gets new models trained with Muse Wonder, Meta's frontier model, while Wonder itself stays reserved for harder subscriber workloads.
Mark Zuckerberg framed the move on X as "bringing Active Memory and Active Focus from Muse Wonder to everyone." Active Memory is meant to carry details between conversations; Active Focus kicks in when a task runs into unexpected problems.
On Meta's internal evaluation, the new Muse models "correctly use critical context in 90% of tasks, up from 76% for the previous model." The current Muse model scores 47% on Meta's Personal Superintelligence Index, up from 38%.
Everything here is company-reported. On the 67% self-correction gain, the article notes Meta "didn't share the baseline." There are no separate numbers for Meta AI, no region or account-tier schedule, and no third-party evaluation. The rollout lands alongside other agent-memory results we've tracked this week.
Originally reported by thenewstack.io
Read the original article →Original headline: Meta Rolls Muse Wonder-Trained Active Memory and Active Focus Models Out to All Muse and Meta AI Users