The ChatGPT update we’ve all been waiting for: AI now talks like a human—it’s amazing.

OpenAI has just unveiled two new AI models, GPT-Live-1 and GPT-Live-1 mini. These two artificial intelligence models are designed to power the voice version of ChatGPT.The models aim to bring voice conversations with ChatGPT closer to a natural exchange with a human interlocutor.

“More than 150 million people use their voice to interact with ChatGPT,” says OpenAI, which hopes to see this number rise thanks to this update.

Better handling of silences

Until now, talking aloud with ChatGPT could be fairly frustrating. Despite its advances, generative AI struggled to string responses together naturally. The assistant regularly interrupted, answered too quickly, or did not give us time to think before speaking. With GPT-Live-1, OpenAI says it has fixed all the flaws inherent in the voice version. The start-up boasts of offering smoother and more natural conversations. For example, it is finally possible to interrupt the AI to ask a question, pause to gather your thoughts mid-sentence, or ask it to slow down.

Introducing GPT-Live, a new generation of voice models for natural human-AI interaction.

Rolling out in ChatGPT starting today.

You’ll want to turn the sound on for this one. pic.twitter.com/WzoQFvA5ir

— OpenAI (@OpenAI) July 8, 2026

To top it all off, ChatGPT now handles silences, which are inevitable during a conversation, better. If the user takes a moment to think, ChatGPT now patiently waits instead of chiming in too quickly or cutting the user off with a follow-up prompt.It was during those moments that conversations with ChatGPT were the most frustrating. The numerous OpenAI demonstrations attended by 01net show that conversations are much smoother thanks to GPT-Live-1. It is almost “like talking to a real person”, explain OpenAI’s engineers.

“GPT-Live is based on a full-duplex architecture, which means it can listen and speak at the same time. During conversations, GPT-Live can show that it is paying attention with expressions such as “mhm” or “yes,” quickly follow on with the exchange, or simply remain silent when you need a moment to think. The result is a voice experience that is surprisingly easy to talk with”, says OpenAI.

Read also: ChatGPT no longer holds the majority of the AI market, a historic first

A smarter voice version

Technically, ChatGPT’s voice version now relies on OpenAI’s latest cutting-edge models, namely GPT-5.6. At launch, however, the AI will make do with GPT-5.5. The startup believes that these models will ensure more relevant responses that are better suited to the context of the conversation. OpenAI says these advances will make it possible to use voice for increasingly complex tasks. GPT-Live-1 is presented as the “most intelligent voice model ever”. Three levels of intelligence will be available in the interface.

It is above all “a much more natural experience”, OpenAI’s engineers say. Nevertheless, the model is not yet perfect. It is the first step toward a truly useful voice assistant capable of meeting a wide range of different needs. This new generation of models brings us “a little closer to a human-AI interaction that feels magical and intuitive”.

Real-time translation

Other use cases include instant translation. As OpenAI has demonstrated, ChatGPT is now capable of translating a conversation in real time, “because it can listen and speak at the same time”. All the world’s most widely spoken languages are supported.

Alongside GPT-Live-1, the AI giant is unveiling GPT-Live-1 mini, a lighter version of the model. It is designed to provide a smoother experience, even on less powerful devices or when the network connection is less reliable. The new models are available to all users on iOS, Android and ChatGPT.com, including users who make do with the free plan. The startup specifies that GPT‑Live‑1 “will become the default model powering ChatGPT Voice for Go, Plus and Pro users, while GPT‑Live‑1 mini will become the default model for users” on the free plan. Deployment of the models has just begun.

A preview of the device designed with Jony Ive?

Could these new voice models foreshadow the release of the mysterious device developed in partnership with Jony Ive, a former Apple designer? According to rumors, the device would rely largely on voice interactions, much like a connected speaker or a smart badge. There is no doubt that the device will need cutting-edge voice models to win users over… According to the latest news, the device has been delayed, and its release is not expected before 2027.

Also read: Robot portrait of OpenAI and Jony Ive’s future AI device – an AI Pin to wear around the neck?

👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.