@finkd
With both models at max effort, Ember matched Claude Haiku 5.5’s Intelligence Index score at roughly one-third the task cost.
The specialized models are designed to retain relevant information and adjust when tasks hit setbacks.
An iOS test described changing Muse’s voice tone by asking the agent or browsing avatar settings.
Separate posts announce real-time Muse conversations, voice options, and phone and video chats.
finkd says the model runs well on new Mac Studios, crediting work with Apple and John Ternus.
Meta says Muse Wonder launched through its Model API and Muse Code, while Full was available to Muse Code subscribers.
CNBC reports the accord calls for internal monitoring, outside assessments and board oversight; Trump called it “morally binding.”
Meta says the platform will initially bring Muse, Meta Business Agent, Muse API and Muse Code to businesses and developers. MongoDB says former CEO Dev Ittycheria has taken over as interim CEO while it searches for a permanent replacement.
Mark Zuckerberg says Meta Enterprise Platform will initially bring the Muse agent, Meta Business Agent, Muse API and Muse Code to businesses and developers.
Meta is also building private processing for certain glasses AI features, which Mark Zuckerberg says will keep that information inaccessible even to the company.
Meta describes Muse as a free personal AI agent for organizing files and handling tasks, with connections to Messages, Calendar and Notes.
Meta says the app works across files, messages, calendars and notes, with users controlling access and the agent always asking before sensitive actions.
Mark Zuckerberg argues customer demand and legal liability give AI labs strong safety incentives. One commenter disputes the liability claim, calling for clearer accountability when AI agents break the law.
Retweet by Meta executive highlights release of updated model from Xiangning Chen.
First streaming audio model from Meta Superintelligence Lab does real-time ASR with diarization.
Meta Superintelligence Labs releases its first real-time audio perception model for streaming speech.
Mark Zuckerberg posted that Meta opened weights for a local model with another release planned soon.
Zuckerberg shares a company essay outlining values for accessible AI development.
Zuckerberg posted that the new model advances Meta toward larger frontier systems with added audit logging.
Zuckerberg posts about terminal agent for full engineering work on large repositories.
