Inworld

The Inworld Realtime API client provides a unified, low-latency interface for building immersive, voice-to-voice multimodal applications. It establishes a single persistent connection to manage speech-to-text, natural turn-taking orchestration, and text-to-speech generation. By utilizing Inworld's multi-model routing system, this client allows developers to stream audio inputs to various underlying language models like GPT, Gemini, or Claude without manually integrating separate microservices or rewriting core connection logic.