OpenAI has launched two new voice models for ChatGPT, GPT-Live-1 and GPT-Live-1 mini, designed to deliver more natural, faster and real-time voice conversations. According to the company, the new models will replace the previous Advanced Voice Mode and provide users with improved voice interaction.
OpenAI has unveiled its new GPT-Live voice technology for ChatGPT with the introduction of GPT-Live-1 and GPT-Live-1 mini. The company said the new models will replace the earlier Advanced Voice Mode. According to OpenAI, the models are designed to provide more natural, faster and human-like conversations while reducing interruptions during AI interactions and maintaining continuous dialogue.
The primary feature of GPT-Live is its Full-Duplex Architecture. Unlike previous voice systems that waited for users to finish speaking before responding, GPT-Live can listen and respond simultaneously, resulting in a more natural conversation flow.
According to the company, the model determines multiple times every second whether it should respond, continue listening, provide brief acknowledgements or use another tool. During conversations, it can also deliver short responses such as "mhmm," "yeah," and "got it," making interactions closer to human conversation.
The company also said that when a query requires search, translation, research or complex reasoning, GPT-Live continues the conversation while completing those tasks in the background using GPT-5.5.
OpenAI stated that GPT-Live-1 and GPT-Live-1 mini performed better than the previous Advanced Voice Mode in human evaluations and technical testing. During conversations lasting five to ten minutes, users preferred the new models over the earlier version. The company's research lead described GPT-Live-1 as its most advanced voice model to date.
The company said the models include additional safety safeguards for responses related to self-harm, violence, mental health and other sensitive topics. Age-appropriate protections and parental controls have also been introduced for teenage users. OpenAI added that GPT-Live does not imitate the voice of any real person and uses only pre-defined voice profiles.
OpenAI confirmed that GPT-Live is being rolled out to all ChatGPT users across iOS, Android and the web. Go, Plus and Pro subscribers will receive GPT-Live-1 by default, while Free users will have access to GPT-Live-1 mini. The company also said the models will become available to developers and enterprise customers through its API in the future.
