Gradium now available for free on LiveKit Inference

Gradium Text-to-Speech is now available on LiveKit Inference, free until Oct. 9.
Gradium TTS is designed for conversational use cases, with voice cloning, low latency (below 250ms TTFA), and robust pronunciation on difficult cases.
LiveKit and LiveKit Inference
LiveKit is an open source framework and developer platform for building voice, video and physical AI agents, on infrastructure that carries billions of calls. For a voice agent it handles everything around the models: realtime transport, turn detection, interruption handling, telephony, and deployment on LiveKit Cloud. LiveKit Inference is the managed way to reach the models themselves from inside LiveKit Agents. You name a model instead of holding an account with each provider, LiveKit runs it on its own infrastructure to keep the round trip short, and usage and rate limits are handled through LiveKit Cloud.
Gradium TTS is one of the models you can name, so an agent already running on LiveKit can move to our voices without a Gradium account, a key or a second bill. The Gradium plugin remains there for teams who would rather manage their own billing and rate limits.
Using Gradium in LiveKit Inference
The model ID is gradium/default, covering English, French, German, Spanish and Portuguese with regional accents inside each language:
from livekit.agents import AgentSession, inference
session = AgentSession(
tts=inference.TTS(
model="gradium/default",
voice="4SZHfMpw-p46Ywgs",
language="en",
),
# ... llm, stt, vad, turn_handling, etc.
)Node.js is supported the same way, and the model is selectable in LiveKit's Agent Builder if you would rather start in the browser.
Text normalization is built in, so the cases voice agents meet constantly, phone numbers, dates, alphanumeric codes and URLs, are spoken the way a person would read them.
Select Gradium as your TTS provider in LiveKit Inference, no additional account or API key required: docs.livekit.io/agents/models/tts/gradium

