Meta says its Muse and Meta AI apps now run on new specialized models trained with Muse Wonder, and two features introduced alongside that work, Active Memory and Active Focus, are now available in both apps.
Muse · Meta AI · Model update
Meta moves Muse and Meta AI onto task-specific models
The apps now run on specialized models trained with Muse Wonder, and two features from that work, Active Memory and Active Focus, are live in both apps. Reported gains are modest in number but the baseline behind them has not been disclosed.
+9%
Personal Superintelligence Index
2
Features now live in both apps: Active Memory and Active Focus
Reported gains, drawn to scale
Each block is 1 percentage point of reported gain. The baseline for both figures is not specified.
+9%
Personal Superintelligence Index
Block heights show relative size only. They do not indicate absolute scores or real-world user impact.
What the figures leave open
Direction is clear. Magnitude is not yet.
What the 9% and 14% gains are measured against: an earlier model or a competitor
Test conditions and how each index is calculated
How many specialized models are involved, and whether they fully replace the existing ones
Rollout scope and regional availability
Independent evaluation of the new models
Why it matters
Meta is building its consumer AI apps around task-specific models, not one general model alone
Active Memory targets retaining and using information across interactions
Both features are already live in Muse and Meta AI
What to watch
Active Focus is not clearly defined in the available material
Meta has not been quoted directly on the update
Gains may look different in everyday use than in the reported indexes
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...