OpenAI has unveiled GPT-Live, an advanced voice AI designed to enhance conversational interactions by allowing simultaneous speaking and listening. This innovation is based on a full-duplex system, enabling the AI to provide responses even as users are still talking. By incorporating natural verbal acknowledgments like “mhmm” or “yeah,” the AI aims to create smoother and more fluid exchanges without noticeable pauses.
In scenarios that demand complex tasks such as web searches or intricate reasoning, GPT-Live efficiently delegates these to a more potent AI model running in the background, ensuring continuity in the user conversation. Currently, these sophisticated tasks are managed by GPT-5.5, with plans to integrate newer models in forthcoming updates.
OpenAI’s rollout of GPT-Live comes on the heels of its announcement regarding the impending public release of the GPT-5.6 model series, which is pending final approval after a thorough cybersecurity review. The GPT-5.6 lineup includes the prominent Sol model, along with its Terra and Luna counterparts.
The introduction of GPT-Live includes two iterations—GPT-Live-1 and GPT-Live-1 mini—that are gradually being made available to ChatGPT users around the globe. Additionally, OpenAI intends to offer GPT-Live through its API, providing developers and businesses the opportunity to integrate real-time voice AI features into their applications.