---
title: "OpenAI Rebuilds ChatGPT's Voice Mode to Listen and Speak at Once"
description: "OpenAI has overhauled ChatGPT's spoken interface with new voice models it says can listen and talk at the same time, aiming for conversations that feel less stilted. The update replaces the company's earlier voice mode and is rolling out to free and paying users."
category: "Technology"
category_url: https://newsparlor.com/category/technology
author: "Noah Andersen"
published: 2026-07-09T01:54:00.000Z
updated: 2026-07-09T01:54:00.000Z
canonical: https://newsparlor.com/article/openai-rebuilds-chatgpt-s-voice-mode-to-listen-and-speak-at-once
tags: ["openai", "chatgpt", "voice-ai", "artificial-intelligence"]
---
# OpenAI Rebuilds ChatGPT's Voice Mode to Listen and Speak at Once

OpenAI has overhauled ChatGPT's spoken interface with new voice models it says can listen and talk at the same time, aiming for conversations that feel less stilted. The update replaces the company's earlier voice mode and is rolling out to free and paying users.

OpenAI has released a new version of ChatGPT's voice feature, built on models it says can listen and speak at the same time, in an attempt to make talking to the chatbot feel more like a normal conversation. The company announced the update this week, and it replaces the previous "Advanced Voice Mode."

## What changed

The central change is technical but easy to feel in use: rather than strictly taking turns, where the user speaks, stops, and then the assistant replies, the new system is designed to handle overlap, so a person can interrupt or react mid-sentence without the exchange grinding to a halt, [TechCrunch reported](https://techcrunch.com/2026/07/08/openai-releases-new-voice-models-for-more-natural-live-conversations/). OpenAI says the models also add small conversational cues, such as brief acknowledgments, and are better at knowing when to pause and let a person think, [The Verge reported](https://www.theverge.com/ai-artificial-intelligence/962856/chatgpt-upgraded-voice-mode-gpt-live). The company said it found testers preferred the new voice to the old one on measures like turn-taking and naturalness; those comparisons come from OpenAI's own testing.

## How it works, and who gets it

For harder questions that need up-to-date information or heavier reasoning, the voice system hands off to one of OpenAI's more capable models in the background and then relays the answer, according to the company's description reported by TechCrunch. OpenAI said the feature is rolling out across its phone apps and the web, with a smaller model set as the default for free users and the fuller version available to paying subscribers. The company did not give a precise date for when the rollout would reach everyone.

## The bigger contest

The update is part of a broader industry bet that talking, rather than typing, will become a more common way people interact with AI. Rivals including Google and Anthropic have been building out their own voice features, and OpenAI's move is a claim to lead on the quality of live conversation specifically. How much of a real-world edge the "listen and speak at once" approach delivers, and whether users adopt voice as a primary way of working with these systems, will become clearer as the new models reach a wide audience and are tested outside the company's own demonstrations.

## Sources

- [OpenAI releases new voice models for more natural live conversations](https://techcrunch.com/2026/07/08/openai-releases-new-voice-models-for-more-natural-live-conversations/)
- [ChatGPT's upgraded voice mode is better at knowing when to stop talking](https://www.theverge.com/ai-artificial-intelligence/962856/chatgpt-upgraded-voice-mode-gpt-live)