FULL STORY

Meta's Muse Realtime Avatar: From Launch to Hands-On

Meta launched Muse Realtime Avatar at Connect with 870ms real-time audio-visual sync. Former researchers frame it as a first step toward fusing LLMs with world models, while rivals like LemonSlice quickly published comparison tests.

2026-09-24 ~ 2026-09-26 · 3 episodes · 23 posts

Episode 1 · Meta Unveils Muse Realtime Avatar at Connect: Real-Time Avatars with ~870ms Latency (2026-09-24, 18 posts)

On September 24, Meta AI released Muse Realtime Avatar, turning the voice stream of Muse Realtime Voice into a conversational digital avatar: animations are generated in sync as it speaks, voice and video stay aligned, first-token response is under one second, and it supports unlimited-length continuous conversation. The news was announced by Alexandr Wang and spread further by Robert Scoble and others.

Confirmed

  • Muse Realtime Avatar pairs with Muse Realtime Voice and can drive the user's Muse digital persona in real time during conversations.
  • Portrait photos can show nuanced expressions; full-body illustrated characters gesture and change poses; animals and everyday objects can also be animated.
  • Response time is under one second, with unlimited-duration continuous conversation supported.

Blind test comparison

  • The Meta team ran a same-character live conversation comparison between Muse Realtime Avatar and Runway Characters and HeyGen LiveAvatar: both sides drove the same character using each product's real-time calling experience, and users preferred Muse. The comparison was organized by Meta itself, so the conclusion is the company's own.

Why it matters

  • A real-time audio-visual-synced digital human goes head-to-head with commercial products like HeyGen and Runway, and it won the blind preference test—signaling Meta's competitive ambitions in this space; sub-second response and unlimited conversation length also lower the engineering bar for real-time virtual human interaction.

Episode 2 · Ex-Meta researcher: real-time interactive video generation to merge LLMs and world models (2026-09-25, 2 posts)

Former Meta researcher Alexandre Conneau says real-time interactive video generation, starting with the newly released Avatars, is merging LLMs with video-generation world models.

Episode 3 · Meta Muse launches real-time AI avatars; LemonSlice fires back with comparison (2026-09-25, 3 posts)

Meta announced real-time interactive AI avatars in Muse, a step toward practical use. Rival LemonSlice quickly posted a comparison video claiming its own real-time avatars hold up against Meta's offering.