Skip to main content

​
Build content that’s alive

Today’s content is offline and passive — you watch, you scroll, and that’s it. Vivix is building the next generation of content infrastructure to turn it into something fundamentally different: interactive, real-time generated experiences. Avatars that perform, listen, and improvise. Worlds and stories that react to what users say, type, or do. The Vivix API gives developers a single platform to generate, orchestrate, and stream this new kind of content — turning viewers into participants, and participants into creators.

​
Core capabilities

Bring characters to life with expressive movement, interaction with their surroundings, and continuous real-time generation.

Full-body motion & lip sync

Create expressive characters with full-body movement and natural lip sync. From speaking and gesturing to singing and dancing, bring large, dynamic actions to life in real time.

Spatial interaction

Let characters move through the scene as the interaction unfolds. Ask them to walk across a room, turn toward you, or approach an object, all guided by your input.

Consistent object references

Bring a product or prop into the scene from a reference image. Keep it in view and visually consistent as the character holds, presents, or interacts with it.

Unlimited streaming

Run live sessions for as long as you want, with no time limit. Keep conversations, performances, and interactive experiences going as your audience shapes what happens next.

​
Use cases

Four examples of what developers can build with Vivix.
c90cf1b4832cb2d0

Customer Support

Engage users in natural, personalized conversations using your knowledge base.
1c5ae7ed82cf69de

Full-Body Live Performance

Create full-body performers that move, interact with their surroundings, and respond live.
7bccb73dcf0a4a58

Interactive Learning

Turn stories and lessons into interactive experiences that respond as the learner participates.
4350423aed47b45b

Interactive Worlds · Preview

Create worlds and stories that react to what users say, type, or do.Watch preview
For a full walkthrough of how these scenarios are built, including the core integration code, see the sample cases: Customer Support and Live Performance.

​
Choose your API

The Vivix API has two product lines. Which one you need comes down to real-time interaction versus asynchronous video generation: A simple test: does your user need to influence the picture while watching it? If yes — the user says something and the avatar responds in real time — choose Streaming Avatar: it delivers a continuous real-time audio and video stream plus a WSS control channel. If not — you just need a video file to play or distribute — choose Blazing Fast Video Generation: it delivers an asynchronous task and a playback URL for the finished video. Both product lines share the same Base URL, the same API key authentication, and the unified {code, message, data} response envelope. See Models for each model’s capability boundaries and selection guidance, including GET /v1/models to list the models enabled for your API key. A third product line, Streaming World (a real-time world model, vivix-w1-stream), is in private beta and not yet publicly available — see Streaming World.

​
Quickstart

Create your account and API key to get started. For Streaming Avatar, try a template next.
1

Create your Vivix account

Register for the Vivix developer platform and sign in to your account.
2

Generate an API key

Open API Keys in the developer platform, create a key, and store it somewhere secure.
3

Start with a template

Building with Streaming Avatar? Run the Avatar Playground template locally with the Vivix Studio CLI to see how an avatar speaks and moves.Open Avatar Playground →
Send REST requests to https://api.vivix.ai with your API key in the request header:
Keep your API key on your server. Any call that sends Authorization: Bearer <VIVIX_API_KEY> — creating or closing a session, submitting a generation — must run on your own backend; never ship your Vivix API key to a browser or mobile app. A browser only needs the values returned by session creation: control.url and control.client_secret for the control channel, plus the TRTC media credentials. control.client_secret is a short-lived, single-session credential: it authenticates one session’s control channel, stays valid only until that session ends (up to max_duration_seconds, 20 minutes by default), and cannot be refreshed — create a new session for a new one. It is not single-use: you can reconnect the control channel with it during the same session, and it stops working the moment the session closes or expires (there is no separate way to revoke it). It is safe to use from the browser; your API key is not. A browser appends it to the returned control URL as the token query parameter; a server-side client may instead send Authorization: Bearer <control.client_secret>. The query parameter name must be token, not client_secret. Do not log the credential or the complete control.url out of logs and analytics as well.
Once you are ready, continue with the product line that matches what you are building:

Continue with Streaming Avatar

For conversation-first, real-time avatar products.

Continue with Blazing Fast Video Generation

For asynchronous video generation.