Meta rolls out Muse Wonder-derived specialized models to Muse and Meta AI: 2.4x less forgetting
mhdfaran · x · 2026-11-07
Meta announced that specialized models trained with its Muse Wonder series are now rolling out to all Muse and Meta AI users, bringing Active Memory and Active Focus to everyone.
Key numbers:
- Personal Superintelligence Index score rises from 38 to 47
- 2.4x reduction in forgetting critical information (Active Memory eval: 76% → 90%)
- 67% better at self-correcting and completing everyday tasks
Active Memory trains models to distinguish temporary vs. permanent info, reason across time, and detect stale context, so users repeat themselves less and hit fewer dead ends.
Related event: Meta Rolls Out Muse Wonder-Powered Model to Muse and Meta AI(4 posts)→
More from Models
- Your LLM Is Secretly a Reward Model: Revisiting Stanford's DPO Paper — dotey · 2026-11-07
- Meta's Muse switches to new specialized models: Active Memory eval jumps 76%→90% — alexandr_wang · 2026-11-07
- Drex Max answers in 183ms median, ~10x faster than GPT-6 Sol in tests — testingcatalog · 2026-11-07
- Astra edit fixes looping by changing just 0.016% of weights; Nasdaq drops 6.3% — eyishazyer · 2026-11-07
- Friday Checks fact-checks the week in AI: GPT-6 Cyber real, China ban is a draft — eyishazyer · 2026-11-07
- Nous launches Portal: one account for 200+ models, bundled tools, and agent cloud hosting — NousResearch · 2026-11-07