Inworld AI
Realtime voice API for consumer apps: TTS-2 text-to-speech, voice cloning, speech-to-speech, STT-1 transcription and LLM inference, priced per character
Overview
Inworld AI sells a realtime voice stack for developers building consumer apps such as companions, language learning, fitness, social and entertainment. The core products are Realtime TTS-2 and the cheaper TTS-2 Flash (streaming text-to-speech with voice cloning and voice design), a realtime speech-to-speech API over WebSocket or WebRTC, the STT-1 transcription model, and OpenAI-compatible inference on open LLMs, with dedicated compute in the US, EU, India and APAC. Inworld started out as an AI character engine for games: its August 2023 round, led by Lightspeed, valued it at over $500 million. Today the homepage leads with voice, and customers it names include Wishroll (Status), Talkpal, Latitude and Bible Chat.
The Verdict
Who Should Use Inworld AI?
Best For
- Consumer apps that need low-cost, high-volume streaming TTS
- Teams building live voice agents who want TTS, STT and LLM calls from one vendor
- Products that need many custom voices or clones per account
- Health or privacy-sensitive apps that need zero data retention or a BAA
Not Ideal For
- Creators who want a no-code studio for voiceovers and audiobooks
- Projects that need the widest set of production-ready languages
- Buyers who rely on a large base of independent reviews
What's Great
- Low list prices: TTS-2 Flash from $15 per million characters on pay-as-you-go, as low as $7 on Growth
- One API covers TTS, speech-to-speech, STT and LLM inference
- Credits roll over for three months; overage bills at plan rates
- Zero data retention from the Developer plan; HIPAA BAA add-on on Growth
- Large voice and clone allowances (30,000 voices and 40 pro clones on Growth)
Watch Out For
- API-first: no studio-style editor for non-developers
- Realtime streaming over WebSocket needs more client work than a simple REST call
- Product focus has moved from game characters to voice, so older guides may be out of date
- Very few independent user reviews; no G2 or Reddit discussion found
Pricing
View all features & details
Key Features
- Realtime TTS-2 and TTS-2 Flash streaming speech
- Voice cloning and voice design
- Realtime speech-to-speech API (WebSocket or WebRTC)
- STT-1 transcription
- OpenAI-compatible LLM inference on open models
- Dedicated deployments in US, EU, India and APAC
Use Cases
- AI companions and social apps
- Language learning and tutoring
- Health and fitness coaching
- Interactive stories and entertainment
How It Compares
| Feature | Inworld AI | ElevenLabs | Cartesia |
|---|---|---|---|
| Focus | Realtime voice for consumer apps | Voice studio plus API | Low-latency voice API |
| Speech-to-speech API | Yes | See vendor | See vendor |
| STT included | Yes (STT-1) | See vendor | See vendor |
| LLM inference | Yes, OpenAI-compatible | See vendor | See vendor |
| Free entry | Pay-as-you-go, no monthly fee | See vendor | See vendor |
Sources: Inworld pricing · Inworld homepage
Is this your tool? Add a badge