AI & Enterprise
OpenAI launches GPT-Live-1 voice AI API for simultaneous listening and speaking
OpenAI has launched its real-time voice conversation model, the GPT-Live-1 API. Unlike existing voice AI that waits for a user to finish speaking, it supports voice agents that can interrupt and switch turns naturally during conversation. OpenAI said the full-duplex approach processes input and output audio in one model, reducing latency and instability. Developers can fine-tune speaking style and offload reasoning to backend models such as GPT-6 Astra.
