Meta AI and Muse Will Run on New Models Trained Using Muse Wonder

What happened

Starting today, Muse and Meta AI will run on new specialized models trained using Muse Wonder. The post describes Active Memory as enabling models to reason across time and remember what matters, and Active Focus as enabling Muse to keep working toward a goal despite roadblocks.

Read next

• More changes in today’s briefingGoogle DeepMind Rolls Out Gemini Omni 1.2 Flash for Longer Video Generation

Reported by 2 sources

The story so far

Meta said Muse and Meta AI are starting to run on specialized models trained with Muse Wonder, adding Active Memory and Active Focus for complex tasks. Meta describes Active Memory as helping the models reason across time and remember what matters, while Active Focus helps Muse keep pursuing a goal when roadblocks arise. In Muse, Meta reports that the updated models reduce forgetting information critical to users’ goals by 2.4x and improve getting unstuck as constraints change by 67%. Meta says the models can also break problems into steps, ask sharper follow-up questions and adapt when users change direction.