[AINews] Meta Connect 2026: Muse glasses, voice, video, and Charm

At Meta Connect 2026, Meta presented Muse, its personal agent, as the center of a hardware-plus-agent strategy, shipping agent features and new glasses and teasing, without releasing, a new frontier model.

Muse now supports voice and real-time video and can hold long conversations while working on background tasks, per Mark Zuckerberg. Video chat with a prompt-customizable voice is marked "coming soon," according to Alexandr Wang. Muse is coming to all Meta glasses, activated by a wake word, also "coming soon." Each Muse gets its own email address for CC and forwarding, and Muse for Mac now performs computer use: "queue up your jobs, walk away, and it keeps going," Wang said.

Wang showed a connector catalog with 1,500+ applications, including Spotify, Box, Temu, Lovable and ElevenLabs. Per a compilation by Claire Jyz, Muse is free for users but Meta may eventually take a cut of transactions; retail and commerce integrations include Walmart, Best Buy, Gap, Sephora and Instacart, with productivity integrations spanning Box, GitHub, Granola and Notion.

Meta also announced Muse Realtime Avatar, a research release that animates a Muse in sync with Muse Realtime Voice, answering in under a second with unbounded session length and output watermarked as AI without adding latency, per Wang and Meta AI. Meta compared it against Runway Characters and HeyGen LiveAvatar using each product's live-call experience; Meta reported Muse "came out ahead on overall preference" and posted no margins or rater counts, with Wang adding "(unsurprisingly)" and details in the research blog.

Hardware included Ray-Ban Meta Gen 3 with longer battery, upgraded microphones and new styles including Aviators; Meta VR Glasses at $1,299, described as Meta's first VR delivered in glasses rather than a headset; glasses turned into an FDA-cleared hearing aid; and Muse Charm, a keychain device shipping in December, per Zuckerberg and Wang.

Meta acquired WaveForms AI, the speech and audio startup led by Alexis Conneau, whose work surfaced at Connect. Wang said "pretty soon we are dropping the most capable model we have ever trained."