OpenAI unveils GPT-Live: ChatGPT can now listen and talk simultaneously

ChatGPT can now listen and talk at the same time, making AI conversations seem more human.

Sam Altman’s OpenAI on July 8, has officially launched GPT-Live, a new generation of voice models designed to replace the existing ChatGPT Advanced Voice Mode.

OpenAI introduced audio models aiming to make voice-based software agents more conversational and capable of completing tasks in real time.

The update introduces two models—GPT-Live-1 for paid Go, Plus, and Pro tiers and GPT-Live-1 mini for free-tier users—rolling out immediately across iOS, Android, and web platforms.

The breakthrough release completely overhauls the underlying framework of AI voice interactions, shifting from rigid, turn-by-turn exchanges to fluid, human-like conversations.

Key features of GPT-Live:

Previous voice modes relied on a "walkie-talkie" cadence, forcing the AI to wait for a clean gap of silence before processing what a user said.

This often led to delayed interruptions caused by background noise or brief thinking pauses while GPT-Live solves this issue via a fundamental engineering shift.

Built on a true full-duplex architecture, GPT-Live continuously processes incoming audio data while simultaneously generating its own spoken response.

Real-Time Decisions:

The model evaluates interaction cues multiple times per second, dynamically deciding whether to keep speaking, pause, listen intently, or process a tool.

Human-Like Interjections:

Because it listens while speaking, the AI can naturally drop in active listening cues without breaking the flow, and it will stay quiet if it detects the user is simply formulating a thought.

"We keep the conversation going while we think in the background."

Research lead for GPT-Live, Kundan Kumar, said, "When GPT-Live has to think hard for a question, it can delegate its reasoning and complex task to GPT-5.5 in parallel, while GPT-Live still remains in conversation with the user."

Seamless weaving with GPT-5:

For heavy tasks requiring deep reasoning, advanced data logic, or live web searches, GPT-Live instantly hands the heavy lifting off to a frontier model GPT-5.5 working silently in the background.

While GPT-5.5 processes the answer, GPT-Live actively keeps chatting with the user, seamlessly weaving the complex answer back into the vocal stream the moment the background model finishes.

Notably, at launch, currently GPT-Live doesn't support voice with video or screen sharing in ChatGPT, though OpenAI says those features are coming soon and the older standard and advanced voice mode with these features will stick around for now.