FINALLY OFFLINE

OpenAI's GPT Live Replaces Voice Mode Starting July 8

By FINALLY OFFLINE | 7/8/2026

Published 27 minutes after the @openai signal was detected.

OpenAI released GPT Live on July 8, 2026, a full duplex voice model for ChatGPT that can listen and speak at the same time. GPT Live 1 mini now replaces Advanced Voice Mode as the free default across iOS, Android and web, while the full GPT Live 1 model and API access roll out alongside it.

Key Points

I open the ChatGPT voice button and for the first time it talks over me on purpose. OpenAI shipped GPT Live on July 8, 2026, a full duplex voice model that listens while it speaks instead of waiting its turn like every assistant before it. It is rolling out now to everyone in ChatGPT on iOS, Android and web, replacing Advanced Voice Mode by default. The pitch is a model that finally behaves like a person on a phone call, not a walkie talkie.

July 8, 2026. Two Models, No Waiting Room.

OpenAI released two GPT Live variants on July 8, 2026, GPT Live 1 and a smaller GPT Live 1 mini, both live inside ChatGPT today. GPT Live 1 mini replaces Advanced Voice Mode as the default experience, while the full GPT Live 1 model is available to try now, with API access for developers coming later through a signup form.

The full duplex architecture is the actual news here. Older voice modes processed one turn at a time, you spoke, it thought, it answered. GPT Live listens and generates output at the same time, which is why OpenAI says it can hold a better sense of timing, jump in with a quick mhmm, and even handle live translation without the usual half second lag that makes translated conversation feel like a phone call with bad reception.

Advanced Voice Mode, the thing GPT Live replaces, launched in 2024 and already felt like a leap at the time. Two years later it reads like a rough draft, still bound by the turn taking structure of a text chat wearing a microphone. Full duplex is the actual architecture shift, not a marketing label.

The Model Delegates the Hard Questions and Comes Back

GPT Live does not try to reason through everything live. When a question needs a web search or deeper reasoning, it hands the task to OpenAI's latest frontier model behind the scenes and folds the answer back into the conversation once it is ready, the same way a smart assistant puts you on a brief hold instead of guessing.

That is the habit change worth watching. Most voice assistants either answer instantly and shallowly, think Siri, or make you wait through a visible loading state. GPT Live tries to keep the conversation moving while the hard part happens off screen, which only works if the handoff is invisible. The first week of real use, not the launch demo, will show whether that seam shows.

Price per problem solved is where this gets interesting. A quick question costs almost nothing to answer live. A question that needs the frontier model behind the scenes costs OpenAI real compute, twice, once to decide it needs help and once to actually reason through it. Bundling that into a free tier is a bet that habit formation now is worth the compute bill later.

Free Today. The Real Cost Shows Up in the Default.

GPT Live 1 mini is free to ChatGPT users everywhere it has launched, no separate subscription required to swap in for Advanced Voice Mode. The lock in here is not a price tag, it is the default setting itself. OpenAI just made voice the path of least resistance inside the app most people already have open on their phone.

That is the same attention economics showing up across Sam Altman's year. His studio just sold a dropped film for 40 million dollars to Neon, a deal covered in our look at the Neon acquisition, around the same stretch that saw AI infrastructure partner SpaceX close its IPO and mint Elon Musk a trillionaire, detailed in our SpaceX IPO coverage. A free, always on voice model is not charity, it is the cheapest way to keep hundreds of millions of people talking to GPT instead of Gemini or Siri.

Try It Before You Compare It to Siri

Try GPT Live if you already use voice mode for anything beyond setting timers, the full duplex back and forth is a real upgrade over the stop and start feel of the old model. Skip the comparison to Siri or Alexa entirely, those assistants were never built to delegate to a frontier reasoning model mid conversation.

The two facts worth remembering are the date, July 8, 2026, and the default, GPT Live 1 mini now replaces Advanced Voice Mode for everyone. Watch the API rollout next. That is when GPT Live stops being a ChatGPT feature and starts being the voice layer other apps build on.

Frequently Asked Questions

What is GPT Live?

GPT Live is OpenAI's new full duplex voice model family for ChatGPT, released July 8, 2026, that can listen and speak at the same time instead of waiting for one side to finish.

When did OpenAI release GPT Live?

OpenAI released GPT Live on July 8, 2026, rolling it out to ChatGPT users on iOS, Android and web the same day.

Is GPT Live free to use?

GPT Live 1 mini is free and now the default voice experience in ChatGPT, replacing Advanced Voice Mode at no extra cost.

What does full duplex mean for GPT Live?

Full duplex means GPT Live can process what a user is saying while it is still speaking, allowing more natural interruptions, timing and live translation.

Does GPT Live replace Advanced Voice Mode?

Yes. GPT Live 1 mini now replaces Advanced Voice Mode as the default voice experience across ChatGPT.

Can GPT Live handle live translation?

OpenAI says GPT Live's full duplex architecture allows it to perform live translation with less lag than the turn based voice mode it replaces.

Is GPT Live available in the API?

Not yet at launch. OpenAI says API access for developers is coming later through a signup form.

How does GPT Live handle complex questions?

For questions that need web search or deeper reasoning, GPT Live delegates the task to OpenAI's frontier model behind the scenes and returns the answer once it is ready.

Topics: artificial-intelligence, spacex, sam-altman, chatgpt, tech, ai-assistants, voice-ai, sam altman, gpt-live, full-duplex, openai

More in tech