OpenAI Introduces GPT-Live for More Natural Voice Conversations
OpenAI has started rolling out GPT-Live, a new voice mode for ChatGPT that aims to deliver more natural, human-like conversations. Unlike previous voice interactions that required users and the AI to take turns speaking, GPT-Live introduces full-duplex communication, allowing ChatGPT to listen and speak at the same time—similar to a normal phone call.
The feature is gradually becoming available to all ChatGPT users, including those on the free plan.
More Fluid, Human-Like Conversations
Previous ChatGPT voice modes operated on a turn-based system, meaning the AI waited for users to finish speaking before responding. While functional, this often resulted in awkward pauses and less natural dialogue.
GPT-Live continuously processes incoming audio while generating responses, enabling it to:
- Listen while speaking
- Respond naturally to interruptions
- Decide when to continue talking or pause
- Deliver smoother, more conversational interactions
These improvements help conversations flow more like those between two people rather than alternating monologues.
Background Web Searches Without Interruptions
GPT-Live also introduces the ability to perform web searches while continuing the conversation.
Previously, ChatGPT would pause speaking while retrieving information online. With GPT-Live, background search agents can gather information without interrupting the dialogue, allowing responses to remain fluid while new information is fetched.
Two Model Variants
OpenAI is releasing GPT-Live in two versions:
- GPT-Live-1
- GPT-Live-1 mini
Although the company has not yet detailed the differences between the models, both utilize the same full-duplex voice architecture.
Part of a Broader AI Voice Evolution
While OpenAI is not the first company to develop full-duplex AI voice technology, GPT-Live is expected to be the first widely deployed implementation available to millions of everyday users.
Several AI startups have already demonstrated similar systems capable of listening, speaking, and performing background tasks simultaneously. GPT-Live brings these capabilities into ChatGPT’s mainstream experience.
Rolling Out Gradually
The rollout begins immediately for ChatGPT users across all subscription tiers, including free accounts. As with most OpenAI feature launches, availability will expand gradually over time, meaning some users may need to wait before GPT-Live appears in their app.
If the technology performs as intended, GPT-Live could represent one of the most significant upgrades to ChatGPT’s voice capabilities, offering faster, more responsive, and more natural conversations than previous voice modes.