What is Inworld AI?
Inworld AI provides speech models and large language model inference for consumer-facing applications. Its APIs cover realtime text-to-speech, speech-to-text, speech-to-speech, and hosted inference for open models. Developers can use these services to add voice conversations, transcription, generated dialogue, narration, and other AI features to products such as companions, learning tools, health and fitness apps, social services, personal agents, games, and entertainment experiences.
The platform includes API documentation and a browser-based playground for testing models. Its infrastructure serves voice workloads from regions in the United States, Europe, India, and Asia Pacific, while the page states that LLM inference is hosted in the United States.
Key Features
- Realtime text-to-speech: Realtime TTS-2 and TTS-2 Flash generate directed synthetic speech, with support for hundreds of languages and dialects.
- Streaming speech recognition: Realtime STT-1 produces transcripts and can identify voice traits such as emotion, style, age, and accent.
- Speech-to-speech sessions: Applications can listen, process a response with a selected LLM, and return spoken audio through one WebSocket or WebRTC connection.
- Open-model inference: Developers can run more than 100 language models through Inworld Inference, with listed models suited to dialogue, roleplay, narrative, classification, and decision tasks.
- Voice creation tools: Available capabilities include custom voices, instant voice cloning, professional voice clones, voice design, and steering for supported TTS models.
- Production controls: Higher plans add increased concurrency, workspace sharing, team management, priority support, compliance options, data residency, and enterprise agreements.
Use Cases
- Add conversational voices to companions, social apps, and personal agents.
- Create narration, character dialogue, and commentary for games and media.
- Build spoken language tutors and other interactive education products.
- Transcribe streaming audio and voice notes with speaker characteristics.
- Connect an application’s chosen LLM to a realtime voice interface.
Pricing
Inworld AI uses credit-based and usage-based pricing. The On-Demand plan starts at $0 with pay-as-you-go rates. Paid monthly plans include Creator at $25, Builder at $100, Developer at $300, and Growth at $1,500; each provides matching monthly credits. Enterprise pricing is custom. Usage rates vary by service and plan, with LLM inference starting from $0.07 per million tokens. Annual billing includes two months free according to the pricing page.
Who Is It For?
Inworld AI is intended for developers, content creators, small teams, and larger organizations building production applications that need generated speech, transcription, conversational voice, or hosted open-model inference.
