OpenAI has unveiled GPT-Live, an innovative voice AI designed to enhance conversational fluidity by enabling simultaneous speaking and listening. This new technology operates on a full-duplex framework, allowing the AI to interject with natural verbal cues like “mhmm” or “yeah,” which contributes to a more seamless and expedited dialogue experience, free from prolonged silences.
For tasks that demand intricate web searches or complex reasoning, GPT-Live can efficiently delegate these to a more advanced AI model running in the background, all while maintaining the user interaction. At the moment, these sophisticated operations are managed by the GPT-5.5 model, but OpenAI plans to incorporate support for newer, cutting-edge models in upcoming updates.
This development coincides with OpenAI’s announcement regarding the public release of its GPT-5.6 model series, pending a thorough cybersecurity evaluation. This series includes the leading Sol model, along with the Terra and Luna variants, marking a significant step forward in AI capabilities.
OpenAI has also started the global rollout of two versions of its new voice models—GPT-Live-1 and GPT-Live-1 mini—to users of the ChatGPT platform. Additionally, the company aims to offer GPT-Live through its API, thus providing developers and businesses with the tools needed to create applications that feature real-time voice AI technology.