ChatGPT Voice Gets a Big Upgrade With New GPT-Live Models

OpenAI is launching today a new GPT-Live family of voice models that will make voice conversations with ChatGPT feel even more natural. The new GPT-Live models allow the chatbot to listen and talk at the same time and show visual results through supported widgets, though there are some current limitations regarding video sharing and connected apps.

“These advances power a new ChatGPT Voice experience that is more intelligent and natural to use. Over time, we believe this research will also unlock the ability to use voice for increasingly complex, longer-running, and more agentic work,” OpenAI explained today.

ChatGPT Voice will now use the GPT-Live-1 model for users on paid plans, and the GPT-Live-1 mini model for free users. These models are already optimized for some of the most popular languages and will be used by default. However, the previous real-time voice experience, which uses turn-based models, remains available for mobile conversations that require video or screen sharing. The new GPT-Live models also don’t support connected apps and plugins yet.

GPT-Live is built on a full-duplex architecture, meaning it can listen and speak at the same time. pic.twitter.com/mPtdpX9cgv

— OpenAI (@OpenAI) July 8, 2026

Unlike OpenAI’s previous turn-based voice models, GPT-Live models are designed for continuous interaction and can process input while generating output. For interactions that require a web search or the use of deeper reasoning, the new GPT-Live models will be able to use OpenAI’s GPT-5.5 models in the background to provide the best results.

The result of these architectural changes should be more natural voice conversations with ChatGPT. “If you take a moment to think, ChatGPT Voice now waits instead of jumping in and interrupting. If you ask it to stay quiet and listen, it will. And when there’s background noise, like passing traffic or nearby conversations, ChatGPT is better at focusing on your voice instead of getting distracted,” OpenAI explained today.

Because ChatGPT’s Voice mode makes the chatbot sound even more human, OpenAI implemented new protections for teens and improved how the chatbot reacts to higher-risk conversations. “We’re also rolling out longer-term measurement and post-launch monitoring focused on emotional reliance to continue improving our understanding and refining safeguards,” the company said today.

The rollout of GPT-Live models in ChatGPT is starting today across iOS, Android, and web, and all free and paid users should get access to the new ChatGPT voice experience over the next few days. Today, the AI startup also announced that GPT-5.6 Sol, its next-generation frontier model that’s currently available in limited preview, will be launching publicly tomorrow along with the smaller GPT-5.6 Terra and Luna models.