OpenAI has unveiled GPT-Live, a cutting-edge voice AI technology designed to enhance conversational interactions by enabling simultaneous speaking and listening. This innovation, built on a full-duplex architecture, allows the AI to interject with natural affirmations like “mhmm” or “yeah,” facilitating smoother and more dynamic exchanges without the typical interruptions or pauses.
For tasks that require more than basic conversational responses, such as web searches or intricate reasoning, GPT-Live can seamlessly delegate these requests to a more robust AI model operating in the background, all while maintaining the dialogue with the user. Currently, these more complex tasks are managed by GPT-5.5, with plans to integrate even newer models in forthcoming updates.
The introduction of GPT-Live comes in the wake of OpenAI’s announcement regarding the forthcoming public release of the GPT-5.6 model series, pending completion of an additional cybersecurity review. The GPT-5.6 lineup will include several models, notably the flagship Sol model, alongside the Terra and Luna variants.
OpenAI is actively deploying two versions of the new voice model, GPT-Live-1 and GPT-Live-1 mini, to ChatGPT users around the globe. Additionally, the company intends to make GPT-Live accessible via its API, allowing developers and businesses to integrate real-time voice AI functionalities into their own applications.