Meta is opening real-time voice and video conversations with Muse, its consumer AI agent, to all users in the United States, the company said Thursday. The capability, branded Muse Voice, pairs a talking animated avatar with new underlying models the company calls Muse Realtime Avatar and Muse Realtime Voice. The feature was first shown at Meta's Connect developer conference in September. Thursday's broader release was announced in a company blog post and in a post on X by Alexandr Wang, Meta's chief AI officer, who shared a demo of his own agent. The default character is a cartoonish figure nicknamed Jolly, and users can swap in a person, an animal or another character. During a conversation, the avatar moves in step with what it is saying, using what the company describes as subtle expressions and gestures timed to the tone of the exchange. Users can adjust how the agent sounds by describing the voice they want and letting the system choose, or by picking from options in the avatar settings. The rollout lands just before Meta's planned launch of a small, Tamagotchi-style wearable for Muse before the end of the year. Real-time conversations are supported on mobile, the web and Mac, with support planned later for Meta's AI glasses and the forthcoming Muse Charm hardware.