
Meta rolls out new models in Muse and Meta AI, with "Active Memory"
Redacción IvokaNov 10, 9:00 AM 3 min
Muse and Meta AI are starting to run on specialized models trained with Muse Wonder, Meta's most capable model series. They bring Active Memory and Active Focus.
Meta is rolling out specialized versions of the models that power Muse (its personal agent app) and Meta AI. According to TestingCatalog, the new models were trained with Muse Wonder, the company's most capable model series, to bring some of that flagship series' capabilities to a wider group of users.
What changes for users:
- Active Memory: lets Muse and Meta AI retain what matters and recognize when a detail from an earlier conversation becomes relevant again. Meta says it trained the models to separate temporary information from permanent information and to identify what remains useful or has become stale.
- Active Focus: is designed to keep Muse working toward a goal when unexpected problems appear. Meta reports a 67% gain in completing everyday tasks where unexpected issues arise.
- In Meta's Active Memory evaluation, the current Muse models correctly use critical context in 90% of tasks, up from 76% for previous models. Meta also cites a 2.4x reduction in forgetting critical information. These are company figures, not an independent test.
In Meta AI, the goal is a stronger thinking partner: breaking complex problems into clear steps, asking sharper follow-up questions, and staying with the conversation when the user changes direction.
Sources
- The New Stack: https://thenewstack.io/meta-muse-specialized-models/
- TestingCatalog: https://www.testingcatalog.com/icymi-meta-rolls-out-specialized-models-for-muse-and-meta-ai/


