OpenAI has introduced GPT‑Live, a new voice interaction mode for its chatbot that supports simultaneous listening and speaking.
Unlike earlier voice implementations that required users and the model to take turns, GPT‑Live operates with a full‑duplex architecture, allowing continuous two‑way communication similar to a phone call.
The technology enables the model to process incoming speech while generating responses, improving its ability to interrupt, pause, or remain silent when appropriate.
In addition to real‑time dialogue, GPT‑Live can launch web‑search agents during a conversation, eliminating the pauses previously required for information retrieval.
The rollout includes two variants, GPT‑Live‑1 and a smaller GPT‑Live‑1 mini, both available to all users, including those on the free tier.
Industry observers note that other AI startups have demonstrated comparable full‑duplex capabilities, but GPT‑Live marks the first widespread deployment of such functionality within a major conversational AI platform.
Early adopters are expected to evaluate the impact of continuous processing on conversation fluidity and overall user experience.
OpenAI plans to gather feedback as the feature becomes active on user accounts, with further refinements anticipated based on real‑world usage.
