Advanced Voice is rolling out to all Plus and Team users in the ChatGPT app over the course of the week.
While you’ve been patiently waiting, we’ve added Custom Instructions, Memory, five new voices, and improved accents.
It can also say “Sorry I’m late” in over 50 languages.pic.twitter.com/APOqqhXtDg
— OpenAI (@OpenAI)August 21, 2025
The company also unveiled five new voices for the chatbot: Arbor, Maple, Sol, Spruce, and Vale (you can listen to them right now). They’ll be available in both Standard and Advanced Voice modes, joining the four voices — Breeze, Juniper, Cove, and Ember — that ChatGPT already offers. OpenAI also noted that, while video and screen sharing are not currently supported in Advanced Voice, those capabilities will roll out at a later date.
What’s more, OpenAI is incorporating a pair of tools to grant Advanced Voice capabilities more in line with the rest of the text-based chatbot experience:memoryandcustom instructions. When it first debuted, Advanced Voice could only reference information from the chat it was currently having. With the memory function, the AI will be able to recall details from previous conversations as well, reducing the need for users to repeat themselves. Similarly, custom instructions are designed to set ground rules for the model to follow when generating its responses. For example, you could dictate that any coding-based responses be presented in Python.
ChatGPT isn’t the only AI that can converse directly with its users. Tuesday’s announcement comes less than a fortnight after Google released its Gemini Liveto all users— including those on the free tier.
