OpenAI unveils GPT-Live to deliver more natural, intelligent voice conversations in ChatGPT

GPT-Live introduces full-duplex voice interactions, live translation, richer visual responses and access to advanced reasoning models, enabling more natural conversations while expanding ChatGPT Voice capabilities for consumers, enterprises and developers across mobile and web.

OpenAI has announced GPT-Live, a new voice model that will power ChatGPT Voice, bringing more natural, responsive and intelligent voice interactions to users. The company said the model represents its most advanced voice technology to date, incorporating research breakthroughs that significantly improve conversational flow and responsiveness.

Unlike earlier voice models, GPT-Live uses a full-duplex architecture, allowing it to listen and speak simultaneously. This enables more fluid back-and-forth conversations, smoother interruptions, improved pacing and a more natural dialogue experience.

Smarter conversations with advanced AI reasoning

GPT-Live is designed to handle more complex interactions by seamlessly accessing OpenAI's latest frontier reasoning models whenever deeper analysis, web search or advanced problem-solving is required. The results are then integrated back into the ongoing voice conversation, allowing users to continue interacting without disrupting the experience.

The enhanced voice model also introduces improved listening capabilities, reducing unnecessary interruptions during pauses while delivering better performance in noisy environments.

Enhanced user experience with visual responses and live translation

The new ChatGPT Voice experience extends beyond spoken conversations by introducing visual response cards for topics such as weather, sports, financial markets and other dynamic information while users remain engaged in voice interactions.

GPT-Live also enables continuous real-time translation, making multilingual conversations more natural by allowing users to switch languages without interrupting the flow of communication.

Rollout begins for ChatGPT users worldwide

OpenAI said more than 150 million people currently use Voice and Dictation features in ChatGPT every week. GPT-Live is now being rolled out globally and will become the default voice model for ChatGPT users.

The company said Go, Plus and Pro subscribers will receive the full GPT-Live experience, while Free-tier users will receive a lightweight version of the model as the default voice assistant. The rollout covers both web and mobile platforms.

Foundation for more capable AI assistants

According to OpenAI, the latest advancements are intended to make voice interactions more intuitive while laying the groundwork for increasingly sophisticated AI assistants capable of handling longer, more complex and agentic tasks over time.

The company also confirmed that GPT-Live will be introduced to its API in the near future, enabling developers to integrate the new voice capabilities into their own applications.