New Introducing $INWORLD, the token for the voice layer of onchain AI. Learn more

The voice layer for onchain AI

Give your agent a voice.

$INWORLD is the token for the voice layer of onchain AI. Meet the host: a realtime AI voice that lives on our pump.fun stream, calls out big trades live, and talks back in 500 languages.

Watch live
Contract address launching soon

Live on stream

The host is on air 24/7. Big buys and sells get a live reaction, written and spoken in realtime by Inworld.

Muted by default. Tap the stream to unmute.

Talk to the host

Speech-to-speech over the Inworld Realtime API. Ask about $INWORLD, the stream, or anything else, in any language. Calls end after 3 minutes.

  • 01Realtime voice in, voice out
  • 02Interrupt any time, like a phone call
  • 03Your mic is only used during the call
Ready

The transcript shows up here once the call starts.

How it works

  1. 01

    Live trades

    The host watches every buy and sell of $INWORLD on pump.fun as it happens and picks the big ones to call out.

  2. 02

    Inworld Inference writes the line

    An open LLM on Inworld Inference turns each event into a short, in-character reaction.

  3. 03

    Inworld TTS-2 speaks it, live on stream

    Expressive, low-latency speech in about 500 languages, straight into the broadcast.

What you can build

Live now

Live AI host

A 24/7 voice host that reacts to big trades on the pump.fun stream.

Next

Voice agent launchpad

Spin up a voiced agent for your own token or community in minutes.

Next

Agents pay per call in $INWORLD

Onchain agents pay for voice and inference per call, no API keys or invoices.

Later

Voice marketplace

Discover, license and use voices built by creators across the network.

$INWORLD token

  • Pay

    Pay for voice

    Use it to pay for voice minutes and agent calls.

  • Hold

    Free Talk minutes

    Holders get free minutes with the Talk voice demo.

  • Fees

    Fees keep the host live

    Creator fees fund the compute that keeps the host on air.

Contract address launching soon
Launching soon

Developers

Read the docs

The host runs on the same APIs you can use today: TTS-2, the Realtime speech-to-speech API, STT and open-LLM Inference. One request gets you a voice.

TTS-2 · curl
curl -X POST https://api.inworld.ai/tts/v1/voice \
  -H "Authorization: Basic $INWORLD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"[excited] gm, Solana.","voiceId":"Dennis","modelId":"inworld-tts-2"}'