Voice-agent and game developers who need a voice that matches conversational tone across 100-plus languages; pay as you go at $25 per million characters.

Kenning's take

What it does

Realtime TTS-2 is a speech model from Inworld AI built for live conversations. It listens to earlier turns of the exchange so it can match the user's tone and pace, takes plain-English directions on how to deliver a line, and keeps the same voice across more than 100 languages. Developers use it through the Inworld API and Realtime API, with a faster TTS-2 Flash version also offered.

• Hears prior conversation audio to match tone
• Voice direction written in plain English
• One voice identity across 100+ languages
• Advanced Voice Design builds a voice from a text description
• Available on REST and Realtime APIs

Who it suits. Developers building voice agents, games, language apps and other real-time voice products.

Key facts

Category

AI models

Part of

Inworld Realtime TTS

Made by

Inworld AI

Released

31 August 2026

How to get it

through the Inworld API and Inworld Realtime API, generally available

Works on

API

Free plan

On-Demand plan is free to start and includes up to 70 minutes of TTS

API price

TTS-2 $25 per million characters on demand, down to $12.50 on Growth; TTS-2 Flash $15 per million

Pricing

Plan	Price	Billed
Creator	$25	month
Builder	$100	month
Developer	$300	month
Growth	$1500	month

source

FAQ

Facts from the maker's own pages and the sources listed. Checked 2026-10-11.