Meta Brings Specialized Models to Muse and Meta AI

Meta is rolling out specialized models trained using Muse Wonder to people who use Muse and Meta AI, extending capabilities from Muse Wonder to the two products. The company says it introduced the Muse Wonder series a few weeks earlier as its most capable models to date for personal and professional agentic tasks.

Meta says the current Muse model scored 47% on its Personal Superintelligence Index, compared with 38% for the previous Muse model. It also says specialized models bring Active Memory and Active Focus, features first introduced with Muse Wonder, to Muse and Meta AI. Active Memory is designed to recall relevant past information and distinguish temporary details from lasting ones.

On Meta’s Active Memory evaluation, the current model correctly used critical context in 90% of tasks, compared with 76% for the previous model; Meta says the new model reduced forgetting of critical information by 2.4x. Active Focus is designed to keep working toward goals when unexpected problems arise. Meta reports a 67% improvement in completing everyday Muse tasks involving unexpected issues.

Meta says the specialized models were trained on real-world bottlenecks to recover, try harder or find alternatives. It says they can manage multiple tasks, decide when to pursue goals independently or collaborate live with users, and adapt to mid-task guidance and clarifications. In Meta AI, the models can break complex problems into steps, ask follow-up questions and adjust when users change direction. Meta says the rollout aims to give more people access to these capabilities.

Meta is rolling out specialized models trained using Muse Wonder to people who use Muse and Meta AI, extending capabilities from Muse Wonder to the two products. The company says it introduced the Muse Wonder series a few weeks earlier as its most capable models to date for personal and professional agentic tasks.

Meta says the current Muse model scored 47% on its Personal Superintelligence Index, compared with 38% for the previous Muse model. It also says specialized models bring Active Memory and Active Focus, features first introduced with Muse Wonder, to Muse and Meta AI. Active Memory is designed to recall relevant past information and distinguish temporary details from lasting ones.

On Meta’s Active Memory evaluation, the current model correctly used critical context in 90% of tasks, compared with 76% for the previous model; Meta says the new model reduced forgetting of critical information by 2.4x. Active Focus is designed to keep working toward goals when unexpected problems arise. Meta reports a 67% improvement in completing everyday Muse tasks involving unexpected issues.

Meta says the specialized models were trained on real-world bottlenecks to recover, try harder or find alternatives. It says they can manage multiple tasks, decide when to pursue goals independently or collaborate live with users, and adapt to mid-task guidance and clarifications. In Meta AI, the models can break complex problems into steps, ask follow-up questions and adjust when users change direction. Meta says the rollout aims to give more people access to these capabilities.