OpenAI Launches GPT-Live: New ChatGPT Voice Model Explained

OpenAI has introduced GPT-Live, a new generation of voice models designed to make talking with AI feel like a real conversation. Announced via LinkedIn, the company calls it its smartest voice model yet — and it’s rolling out in ChatGPT starting now.

The Headline Feature: Full-Duplex Voice

The biggest technical shift in GPT-Live is its full-duplex architecture. In plain terms, the model can listen and speak at the same time — the way two people do in a normal conversation.

According to OpenAI, this unlocks three things:

More natural back-and-forth — the conversation flows instead of following a rigid “you talk, then it talks” pattern.
A better sense of time — the model keeps track of the rhythm of the conversation.
Live translation — the simultaneous listen-and-speak capability makes real-time translation possible.

Built to Be Enjoyable to Talk To

Beyond the architecture, OpenAI is positioning GPT-Live as a genuinely good conversationalist. The company says a voice model should be enjoyable to talk to, and describes GPT-Live as having a more natural and defined personality than its previous voice models.

Smart Delegation Behind the Scenes

GPT-Live isn’t limited to what it can answer in the moment. For questions that require web search, deeper reasoning, or more complex work, the model can delegate the task to OpenAI’s latest frontier model behind the scenes — then bring the result back into the voice conversation once it’s ready.

In practice, that means the voice conversation stays fluid while heavier work happens in the background, rather than the voice model hitting a wall on harder questions.

Availability: Who Gets It and How

The rollout is broad from day one:

Rolling out to everyone in ChatGPT — across iOS, Android, and web.
API access is coming soon, per OpenAI’s announcement.
How to try it: just tap the Voice button in ChatGPT, and you’ll get the new experience powered by GPT-Live.

FAQ

What is GPT-Live? GPT-Live is OpenAI’s new generation of voice models for natural human-AI interaction, described by the company as its smartest voice model yet. It’s rolling out in ChatGPT now.

What does “full-duplex” mean in GPT-Live? Full-duplex means the model can listen and speak at the same time, enabling more natural back-and-forth conversation, a better sense of time, and live translation.

How do I use GPT-Live in ChatGPT? Tap the Voice button in ChatGPT on iOS, Android, or web. The new voice experience powered by GPT-Live is rolling out to everyone.

Is GPT-Live available in the API? Not yet — OpenAI says API access is coming soon.

Can GPT-Live search the web or handle complex questions? Yes, indirectly. For questions needing web search, deeper reasoning, or more complex work, GPT-Live delegates to OpenAI’s latest frontier model behind the scenes and returns the result within the conversation.

Source: OpenAI’s GPT-Live announcement, shared via the company’s official LinkedIn page.