Developers

Lookalike API

The Lookalike API. A real face for anything you build.

Give any app, agent, or website a hyper-realistic AI video human. Send text, get back a synchronized voice and face streaming in real time. One REST call to start, one LiveKit connection to render, nothing to host.

Free to start · No credit card · Model and framework agnostic

POST /v1/sessions/start
curl -X POST https://api.lookalike.com/v1/sessions/start \
  -H "Authorization: Bearer $LOOKALIKE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "identity_id": "your-identity-id" }'
200 OK
{
  "room_name": "rm_8f3c…",
  "token": "eyJhbGciOi…",        // pre-minted LiveKit JWT
  "livekit_url": "wss://…",
  "dispatch_id": "dsp_51b2…",
  "created_at": "2026-07-06T18:04:11Z"
}

How it works

Three steps to a live face

From API key to talking digital human

STEP 01 / 03

Create an identity

One identity_id bundles a face, a cloned voice, and a persona. Build one from a photo and a voice sample, or start from a preconfigured identity.

STEP 02 / 03

Start a session

POST /v1/sessions/start with your bearer API key. The response returns room_name, a pre-minted LiveKit token, and livekit_url. No token signing on your side.

STEP 03 / 03

Stream the conversation

Connect with any LiveKit SDK. Text you send as assistant_text comes back as assistant_voice and assistant_video with sub-second latency.

Connect with any LiveKit client
import { Room } from 'livekit-client'

const room = new Room()
await room.connect(livekit_url, token)
// send your agent's text; it returns as voice + video
// assistant_text -> assistant_voice + assistant_video

Why Lookalike

The easiest way to give AI a face

Everything hosted, nothing to rig

Hyper-realistic video humans

Lifelike faces built from a single photo, not cartoon avatars or 3D rigs. The same engine behind Lookalike's digital humans.

Sub-second latency

Realtime streaming over LiveKit, fast enough for natural face-to-face conversation instead of pre-rendered clips.

Fully hosted gateway

Face rendering, voice synthesis, and video transport run on our infrastructure. You operate no GPUs, media servers, or avatar pipeline.

Model agnostic

Bring any LLM: OpenAI, Anthropic, Gemini, or your own. Your agent thinks, the Lookalike API gives it a voice and a face.

Text, voice, and video modes

Serve the same identity across chat, calls, and realtime video from a single gateway and a single identity_id.

A no-code path too

Not ready for the API? Embed any Lookalike with one iframe and upgrade to the API when you need programmatic control.

Full reference, OpenAPI spec included

Endpoint schemas, session options like avatar_enabled and agent_type, and the complete OpenAPI artifact live in the developer documentation.

FAQ

Lookalike API questions

Questions developers ask

What is the Lookalike API?

The Lookalike API is a REST and realtime API for adding a hyper-realistic AI video human to any app, website, or AI agent. You send text (assistant_text) and Lookalike streams back the same content spoken aloud (assistant_voice) on a lifelike talking face (assistant_video), in real time over LiveKit.

How do I add an AI avatar to my app with an API?

Call POST https://api.lookalike.com/v1/sessions/start with your API key and an identity_id. The response contains LiveKit credentials (room_name, token, livekit_url). Connect with any LiveKit SDK and your avatar is live: text you send comes back as synchronized voice and video.

Does the Lookalike API work with ChatGPT, Claude, or Gemini agents?

Yes. The API is model agnostic. Your agent produces text with any LLM, you forward that text as assistant_text, and Lookalike renders the face and voice. It works the same with OpenAI, Anthropic, Google, or a local model.

What is an identity_id?

One identity_id bundles a face, a cloned voice, and a persona. Create identities from a photo and a voice sample, or start from a preconfigured one, then pass the id when starting a session to render that digital human.

How fast is the video? What is the latency?

Sessions stream over LiveKit with sub-second latency, fast enough for natural, face-to-face conversation rather than a rendered clip that arrives seconds later.

Do I need my own streaming or avatar infrastructure?

No. Lookalike is fully hosted: the face rendering, voice synthesis, and realtime video transport all run on our gateway. You make one REST call and connect a LiveKit client. There is no avatar pipeline, 3D rig, GPU fleet, or media server to operate.

Is there a no-code alternative to the API?

Yes. Every Lookalike can be embedded with a single iframe (https://lookalike.com/embed/YOUR_TOKEN?modes=video), which is the fastest path for websites and site builders. The API is for products and agents that want programmatic control.

How much does the Lookalike API cost?

You can create a Lookalike and start building for free, no credit card required. Partner API access is provisioned with an API key; see the pricing page for plans or contact us for volume pricing.

Ship a face, not a text box

Start with the docs, embed a Lookalike in minutes, or wire the realtime API into your agent. Your users will never talk to a blinking cursor again.