OpenAI launched GPT-Live: ChatGPT now listens and speaks simultaneously

On July 8, 2026, OpenAI launched GPT-Live — a new generation of voice models for the ChatGPT platform. The main innovation: full-duplex communication — the model can listen and speak simultaneously without requiring pauses between user utterances.

How it works

In the previous version of ChatGPT's voice interfaces, the model used half-duplex mode: the user speaks, the model waits for a pause in speech, then provides an answer. This created an unnaturally feeling. GPT-Live eliminates this limitation: the model processes the incoming audio stream in real time and can respond without waiting for the user to finish their sentence.

This brings interaction closer to natural human conversation, where people often start responding before the speaker finishes their sentence.

Two versions of models

OpenAI released two versions:

• GPT-Live-1 — full-featured model with all ChatGPT capabilities
• GPT-Live-1 mini — lightweight version for mobile devices and devices with limited resources

Both versions are available to all ChatGPT users worldwide from the moment of launch (July 8, 2026).

What it means for conversational AI

Full-duplex communication is a meaningful step toward natural interaction with AI. Until now, most voice assistants (Siri, Alexa, Google Assistant) operated in half-duplex mode. OpenAI demonstrates that full-duplex technology is technically feasible on large language models.

This also shows that companies are investing in improving conversational interfaces — likely because voice is becoming the primary method of interaction with AI on mobile devices.

Why it matters

A mobile AI assistant that talks like a human (without awkward pauses) can attract millions of users. This is especially relevant for markets where text interface is less convenient (developing countries, elderly people, multitasking).

CPU/GPU requirements for full-duplex processing are higher than for half-duplex, but the lightweight GPT-Live-1 mini version allows deployment of the model on mobile devices.

What it means

Full-duplex communication in ChatGPT is an intermediate step toward personal AI assistants that function like natural conversationalists. The industry is moving toward reducing friction between humans and models. OpenAI shows that the boundary between "talking machine" and "smart assistant" is blurring.