Meta Launches Real-Time Video Chat with Muse AI Avatars for US Users
Meta has released Muse Voice in the U.S., allowing users to video chat with customizable AI avatars powered by new real-time voice and avatar models.
Meta has officially rolled out video chat capabilities for its consumer assistant, Muse, making the feature available to all users across the United States. First previewed at the Meta Connect developer event in September, the update allows people to hold live voice and video conversations with an animated avatar directly on their devices.
The rollout was announced through a company blog post and a social media update on X by Meta’s chief AI officer, Alexandr Wang. The updated interface relies on two underlying technologies: Muse Realtime Avatar and Muse Realtime Voice models. Together, these systems generate interactive visual agents that respond dynamically to voice input.
Interactive avatars and hardware expansion
Unlike competing AI assistants that rely solely on audio output, Meta’s Muse incorporates a visual identity. By default, users interact with a cute, animated avatar named Jolly, though the character can be customized to look like a human, an animal, or a different character type entirely.
During the announcement, Wang published a demonstration video featuring a customized fluffy white dog avatar. In the demo, Wang joked about his penchant for posting memes on X and how it might cause issues with his boss. The avatar responded by suggesting he add disclaimers to posts that could be viewed as risky or endorsement-related.
Meta explained that as the avatar speaks, it utilizes subtle facial expressions and body gestures tailored to match the overall tone of the conversation. The company believes this form of anthropomorphism makes interactions feel more natural, potentially fostering a stronger connection between users and their digital assistants over time.
The feature is currently accessible on mobile devices, web browsers, and Mac computers. Meta plans to bring the real-time conversation technology to its AI Glasses as well as an upcoming hardware device called the "Muse Charm." Expected to launch before the end of the year, the Tamagotchi-like wearable will feature a small screen displaying the digital avatar, allowing users to talk to the agent and have it respond or perform tasks on their behalf.
Strategy behind visual conversational AI
Meta’s deployment of visual AI agents represents a distinct direction in consumer AI interface design. While several major tech companies have released low-latency voice capabilities for their models, Meta is placing a strong emphasis on visual representation alongside voice interaction.
By pairing real-time audio generation with responsive graphical avatars, Meta aims to transform routine assistant interactions into more engaging, personal conversations. The ability to switch avatar models—ranging from stylized characters to realistic representations—is designed to adapt the interface to individual user preferences.
What it means for developers
The introduction of Muse Voice highlights key industry shifts toward multimodal user interfaces that combine low-latency speech processing with real-time visual output. For developers working on AI-driven software, Meta's implementation offers several relevant takeaways:
- Multimodal integration: Combining real-time voice synthesis with visual animation requires strict synchronization between audio streams, text generation, and dynamic graphics. Developers building customer-facing bots or virtual assistants may need to consider how visual feedback alters user retention and satisfaction.
- Cross-platform deployment: Meta’s strategy targets a broad hardware footprint, spanning desktop software (Mac), web applications, mobile platforms, wearable devices (Muse Charm), and smart glasses. Building agents that function consistently across different screen sizes and display types is becoming a central technical requirement.
- Customizable agent personae: Allowing users to alter the appearance and style of an AI agent suggests a demand for personalized consumer experiences rather than static bot interfaces.
As multimodal interfaces evolve, developers looking to integrate conversational intelligence into their own products need reliable access to underlying LLMs. Developers can try top AI models cheaply through one API at https://apixoai.online to test and compare performance across leading options like GPT-4o, Claude, Gemini, and DeepSeek.
Meta’s upcoming launch of the Muse Charm wearable before year-end will further test whether consumers are ready to carry dedicated visual AI hardware in their daily lives.
Source: Meta says US users can now video chat with their Muse AI avatar — TechCrunch AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
