Build more natural voice experiences with GPT‑Live‑1 in the API
OpenAI has released the GPT-Live-1 model to the API, enabling full-duplex voice interaction capabilities. The update introduces native support for custom voice selection and direct telephony integration.
Verified State Diff
Impact & Verification Analysis
Developers building real-time voice agents, customer service automation platforms, and telephony-based AI applications.
This reduces architectural complexity for real-time voice applications by eliminating the need for external streaming infrastructure and provides a competitive edge in latency-sensitive conversational AI.
Full Fact Overview
The introduction of GPT-Live-1 marks a transition from standard request-response voice processing to a full-duplex architecture, allowing for simultaneous audio input and output. By exposing this via the API, OpenAI is enabling developers to bypass third-party middleware for low-latency voice applications. The inclusion of telephony support suggests a specialized integration layer for PSTN (Public Switched Telephone Network) connectivity, while the custom voice feature implies access to a broader set of synthetic voice profiles beyond the standard TTS offerings.