OpenAI has unveiled GPT-Live, a cutting-edge voice AI system engineered to enhance the natural flow of conversations by enabling simultaneous speaking and listening. This innovation is based on a full-duplex architecture, allowing the AI to interject with natural conversational cues like “mhmm” or “yeah” while users are still talking. The aim is to create smoother and quicker interactions by eliminating lengthy pauses.
For more intricate tasks requiring web searches or advanced reasoning, GPT-Live can seamlessly delegate these to a more robust AI model operating in the background, all while maintaining the conversation with the user. These tasks are initially powered by the GPT-5.5 model, with plans to incorporate newer models in upcoming updates. The launch of GPT-Live comes shortly after OpenAI announced that its GPT-5.6 model series is set for public release, pending final cybersecurity evaluations. This series features the flagship Sol model, alongside Terra and Luna variants.
The rollout of the new voice models, GPT-Live-1 and GPT-Live-1 mini, has begun for ChatGPT users around the world. OpenAI is also preparing to offer GPT-Live through its API, which will enable developers and businesses to integrate real-time voice AI capabilities into their applications. This move signifies a major step forward in making conversational AI more accessible and versatile for various use cases.