Meta brings Active Memory and Active Focus to specialized Muse and Meta AI models
On November 6, Meta said specialized models trained with Muse Wonder are rolling out to Muse and Meta AI. It reports the Active Memory eval rising from 76% to 90%, about a 2.4x drop in forgetting critical information, a 67% gain on everyday tasks when surprises hit under Active Focus, and the Personal Superintelligence Index moving from 38% to 47%. My take: for assistants, memory and goal-sticking matter more than parameter bragging. Before you buy the newest model for work chatbots, ask whether it forgets the order and whether it can reroute when something breaks.