The decision follows the growth in the use of the new technology. Since the arrival of GPT-Live, the number of users conversing with ChatGPT via voice has almost doubled. Additionally, almost all users who use audio features have already migrated to the model. With the closure of previous options, OpenAI aims to focus efforts on developing a single voice experience.
GPT-Live also incorporates learnings from the use of Advanced Mode and Standard Voice. According to OpenAI, this information helps improve the depth of responses, the continuity of context between conversations, and personalization. The proposal is to integrate voice interaction into the rest of the experience of ChatGPT, with more natural conversations and responses that maintain context.
One of the differences of GPT-Live is in the ability to listen and speak at the same time. The model can respond to interruptions, use interjections like "mhmm" and "yeah" to indicate it is following the conversation, and respect pauses while the user thinks. It can also delegate complex tasks to more advanced text models that work in the background.
The previous versions had limitations in this type of interaction. The Advanced Voice operated with stricter turn-taking and needed to wait for the user to finish speaking before responding. The Standard Voice chained three separate systems — speech-to-text conversion, language model processing, and text-to-speech conversion —, which increased response time and could cause loss of conversation nuances.
It's worth noting that previous conversations won't be lost with the change. Chats made with Advanced Voice and Standard Voice will continue in the ChatGPT history and can be resumed via text or voice. In the latter case, the interaction will be handled automatically by GPT-Live.
How will the transition process work?
The transition begins on November 10, 2026, when users will receive notifications about the discontinuation of the old modes. Until December 11, both options will remain available normally. After this date, they will be removed from the selector in Settings > Voice, and users who continue using these versions will be automatically migrated to GPT-Live.
Live video and screen sharing will continue to work with the Advanced Mode until GPT-Live receives native support for these features. However, OpenAI has not yet announced a date or forecast for this change.
If you liked the content, you might also be interested in checking out "7 tips for using the new Hyper-realistic Voice Mode of ChatGPT".
