This is the new ChatGPT Voice, powered by GPT-Live

YouTube AI Channels Products

Summary

OpenAI has released the new generation ChatGPT voice feature powered by GPT-Live, enabling full-duplex real-time conversations, supporting interruption, intelligent reasoning, web search, and real-time translation, making interactions more natural and intelligent.

No content available
Original Article
View Cached Full Text

Cached at: 07/09/26, 03:13 PM

TL;DR: OpenAI releases next-gen ChatGPT voice feature, powered by GPT-Live, enabling full-duplex real-time conversations with support for interruptions, reasoning, web search, and live translation. ## Next-Gen ChatGPT Voice: More Natural, More Intelligent Today, OpenAI released a new ChatGPT voice feature powered by **GPT-Live**, a full-duplex conversational partner that is being called the most powerful voice model yet. ### What is a Full-Duplex Voice Model? In the demo, the model was asked to explain "full-duplex voice model" in simple terms: > "Imagine a normal phone call with a friend—you can listen and speak at the same time. That's full-duplex." This technology makes voice interactions feel much closer to real human conversation—handling interruptions, pauses, corrections, and the natural flow of thinking while speaking. ### Intelligent Reasoning & Real-Time Web Search The new voice doesn't just converse naturally; it can also reason and search the web for the latest information. In the demo, the user is giving a lecture on the history of sound in 30 minutes and needs to verify a few key dates: - Scott de Martinville's phonautograph (1857) - Edison's tinfoil phonograph (1865) - Berliner's gramophone (1887) At the same time, the user also asks to: - Check if the 16th Street BART station has delays - See if it will rain in the afternoon - Pull any current announcements from Committee 60 The model processes all requests in parallel. Ultimately, it corrects a critical error: > "Hmm, one thing to correct. Edison's tinfoil phonograph should be **1877**, not 1865." The user laughs and responds, "You know what? I actually knew that—I was just testing you." This demonstrates that the model not only knows the answers but also proactively identifies and corrects the user's mistakes—a leap in intelligence. ### Live Translation Demo ChatGPT Voice also brings intelligence to real-time translation. The presenter invites a friend, SJ, to role-play as an antiquarian bookseller in Paris, and enables live French translation: - "I love this book." → "J'aime ce livre." - "This is my beloved book." → "C'est mon livre préféré." - "But I can't pay you a high price because this book is full of annotations." → "...le prix fort parce que le livre est tout annoté." - "Tell her $30, and we have a deal." → "Et on conclut. Oh, bon madame, bon. Merci beaucoup." The model smoothly alternates between listening, translating, and speaking the target language, without any waiting—like simultaneous interpretation. ### More Natural Interactions & Exits Throughout the conversation, the model knows when to step in and when to step back. When the user says "stop," it stops immediately and naturally transitions to the next topic (restaurant recommendations). This level of intelligence and naturalness makes ChatGPT Voice a true conversational partner, not just a question-answering tool. --- Source: This is the new ChatGPT Voice, powered by GPT-Live - YouTube (https://www.youtube.com/watch?v=EAN5Cj347PY)

Similar Articles

Listening & Speaking with GPT-Live

YouTube AI Channels

OpenAI demonstrates the real-time simultaneous interpretation and conversation capability of the new voice model GPT-Live, which can listen and speak simultaneously and seamlessly switch between translation and chatting.

Natural Conversations with GPT-Live

YouTube AI Channels

OpenAI showcases breakthrough in personalization and naturalness of the new voice model, capable of natural brainstorming conversations, displaying empathy and timely interjection, approaching human conversation rhythm.

Introducing GPT‑Live

Simon Willison's Blog

OpenAI has finally upgraded the model powering ChatGPT's voice mode with GPT-Live, which can delegate complex tasks to GPT-5.5 while maintaining conversation flow. The new model is significantly more capable than the previous GPT-4o-based voice mode.

GPT‑Live

Hacker News Top

OpenAI announces GPT-Live, a new full-duplex voice model that enables more natural, real-time conversations by allowing simultaneous listening and speaking, with GPT-5.5 as the backend model.