In its announcement, OpenAI described GPT-Live as a new generation of voice models designed to make conversations with AI feel much more like talking to a real person.
Unlike traditional voice assistants that wait for users to finish speaking before responding, GPT-Live is built on a full-duplex architecture, enabling it to listen and speak simultaneously.
Explaining how the new model transforms conversations, OpenAI said GPT-Live can signal that it is actively listening with natural acknowledgements such as "mhmm" or "yeah," engage in rapid back-and-forth exchanges, or remain silent when users need a moment to think.
"The result is a voice experience that is refreshingly easy to talk to," the company said.
OpenAI also described GPT-Live as its smartest voice model yet. For tasks that require web searches, advanced reasoning or more complex problem-solving, GPT-Live seamlessly hands them off to OpenAI's latest frontier model in the background before returning the results to the ongoing conversation.
"While it works, GPT-Live can keep talking with you and maintain the flow of conversation. At launch, GPT-Live will use GPT-5.5 in the background. As we release new frontier models, we'll continuously update the model used by GPT-Live," the company said.
According to OpenAI, these improvements power a new ChatGPT Voice experience that is more intelligent, responsive and natural.
"These advances power a new ChatGPT Voice experience that is more intelligent and natural to use," the company said.
Looking ahead, OpenAI said it believes the underlying research behind GPT-Live will enable voice-based AI assistants to handle increasingly complex, long-running and more autonomous tasks.
The company has begun rolling out two versions of the new model—GPT-Live-1 and GPT-Live-1 mini—to ChatGPT users worldwide.