Physix Frontier · News Briefing Card (IT Home · Nov 7, 2026)
Meta brings Muse Wonder active memory and focus to existing models
KEY FACTS
- Meta on November 6 released a dedicated version of its existing model trained on Muse Wonder.
- The new version has been fully rolled out to Muse and Meta AI users.
- The active memory and active focus features previously debuted with Muse Wonder.
- The new Muse model cuts the problem of forgetting key information by 2.4x.
- When unexpected events occur in everyday tasks, the new Muse model improves task completion rate by 67%.
KEY DATA
2.4xReduction in forgetting key information
67%Task completion rate gain under unexpected conditions
90% (previously 76%)Share of cases where Muse correctly uses key context
47% (previously 38%)Personal superintelligence index score
PHYSIX OBSERVATION
Meta is pushing the memory and focus capabilities of its flagship model down to its everyday assistant, aiming to retain users through the experience gap. But all the improvement figures come from internal evaluations and lack third-party verification, so the real-world effect is questionable. For users, subscribers get the new model first, and capability tiering is becoming the norm in the commercialization of AI assistants.
Source: IT Home report
Physix Frontier