Voice-agent and game developers get fast, expressive speech with cloning at $25 per million characters, falling on the Growth plan.

Kenning's take

What it does

Inworld Realtime TTS-2 is a speech model and API for real-time conversation, with a median time to first audio under 200 milliseconds for the speech layer. It keeps one voice across more than 100 languages, takes free-text directions on tone, supports tags such as laughs and sighs, and can clone a voice from a 5 to 15 second sample. It is sold per character through the Inworld API and is also listed on platforms such as Cloudflare and LiveKit.

• Under 200ms median time to first audio
• One voice across 100+ languages
• Free-text voice direction and non-verbal tags
• Voice cloning from a 5-15 second sample
• Voice design from a text prompt

Who it suits. Developers building voice agents, games and apps that need fast, expressive speech.

Key facts

Category

Developer tools

Made by

Inworld AI

Released

31 August 2026

How to get it

self-serve through the Inworld API and Realtime API, with a browser playground; free to start

Works on

API, Web

Free plan

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

API price

TTS-2: $25 per million characters on demand ($20 Creator, $17.50 Builder, $15 Developer, $12.50 Growth); TTS-2 Flash: $15 on demand down to $7

Pricing

Plan	Price	Billed
On-Demand	$0	month
Creator	$25	month
Builder	$100	month
Developer	$300	month
Growth	$1500	month
Enterprise	Custom pricing	month

source

Models and features

• Inworld Realtime TTS-2 2026-08-31

FAQ

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