OpenAI has released GPT-Live-1 in its API, extending the full-duplex voice model already used in ChatGPT to developers building voice applications and business workflows. The model can listen and speak at the same time, handle interruptions and brief acknowledgments, and continue the conversation while deeper reasoning or actions are delegated to a separate backend model and agent stack.
The architecture matters because GPT-Live-1 is positioned as the conversational front end rather than a replacement for the reasoning system behind an agent. Developers choose the backend model, tools and agent framework, while GPT-Live-1 handles the live voice interaction. OpenAI also lists WebRTC, WebSocket and telephony support, plus transcripts, turn detection and keyword biasing for production voice systems.
OpenAI has expanded the available voice selection across accents, dialects and languages. The company prices GPT-Live-1 at $0.05 per minute for the front-end voice layer; use of the paired backend model and other services is separate.
AiToolMap treats performance claims in OpenAI’s launch material as vendor-reported unless independently reproduced. The concrete product change is availability: GPT-Live-1 is now offered through the API, giving developers direct access to OpenAI’s simultaneous listening-and-speaking voice layer.