Inworld AI provides realtime speech and language tools for consumer-facing applications, with API, self-hosted, and web options. Its TTS-2 supports voice direction, voice cloning, and 200+ languages; the homepage states TTS-2 reaches its first byte in under 100 ms. The Realtime API carries speech in and out through one WebSocket and supports custom voices and tool calling. Realtime STT handles streaming transcription, timestamps, diarization, custom vocabulary, and voice signals such as emotion, age, accent, pitch, and style. Realtime Router can send requests to OpenAI, Anthropic, Google, and 220+ models, with failover, A/B testing, and analytics. Inworld also lists OpenAI compatibility, Twilio integration for TTS, an Unreal SDK, and a Unity SDK in early access. Its plans range from free usage-based On-Demand to paid plans from $25/mo; paid plans provide monthly credits equal to the plan price. It lists SOC 2 Type II certification and HIPAA and GDPR compliance. Zero data retention is available by request and listed as a Growth add-on.
Who it is for
It suits teams building consumer-facing applications in areas such as games and media, education, wellness, or social and companion experiences. Developers needing realtime speech APIs, voice cloning, or LLM routing may also find its listed tools relevant.
What is good
- TTS-2 supports voice cloning and 200+ languages.
- Realtime API combines speech input and output over one WebSocket.
- STT includes timestamps, diarization, and custom vocabulary.
- Router supports 220+ models and failover.
- Free On-Demand plan is available.
What to know first
- Paid plans start at $25/mo.
- Unity AI Runtime SDK is in early access.
- Zero data retention is an add-on for Growth.
Verdict
Inworld AI brings realtime speech, transcription, and model routing into a set of API, web, and self-hosted offerings. The free usage-based option provides an entry point, while advanced usage and support are tied to paid plans.
Inworld AI plans and pricing
All plansCompared on game voice generators
- Free plan
- Yes
- Paid from
- $25/mo
- Commercial use
- Yes
- Voice cloning
- Yes
- API access
- Yes
- SSML or speech controls
- Yes
- Language count
- 200 languages
- Audio export formats
- MP3, WAV, PCM, LINEAR16, OGG_OPUS, μ-law, A-law, FLAC
- Game engine tools
- Unreal AI Runtime SDK; Unity AI Runtime SDK early access
