OpenAI is integrating Bidi 1 into ChatGPT — a voice model that speaks and listens at the same time

OpenAI is accelerating the transformation of ChatGPT into a super app: the company is simultaneously developing a bidirectional voice model Bidi 1, embedding a programmer assistant Codex, and expanding agentic capabilities.

What is GPT Bidi 1

In the code of the ChatGPT mobile app, researchers found traces of a new audio model with the code name GPT Bidi 1. "Bidi" stands for bidirectional. This is a fundamentally different approach compared to the current voice mode, where the model sequentially switches between listening and responding: first it listens, then it speaks — like a walkie-talkie with a "push-to-talk" button.

Duplex mode changes the very dynamics of the dialogue. Users will be able to interrupt the assistant mid-sentence, and the assistant will react to intonation, pauses, and speech pace in real time, without waiting for a pause. The delay between exchanges is noticeably reduced, and the conversation becomes less robotic.

This is how the most advanced voice interfaces work: Apple Intelligence in iOS 18 and Google Duplex have long supported full duplex exchange.

What changes with Bidi 1:

• voice and hearing work simultaneously — without waiting for the "airtime"
• user can interrupt the response at any moment
• response time between exchanges is reduced
• the assistant hears intonation and reacts to pauses
• conversation sounds more natural and closer to dialogue with a human

The specific release date for Bidi 1 has not been announced yet, but the presence of the model in ChatGPT's production code suggests that the release is already near.

Codex inside ChatGPT

In parallel with voice development, OpenAI is embedding its programmer assistant Codex into ChatGPT. Until now, Codex existed as a separate product with its own interface. After integration, you will be able to write, debug, and explain code directly in the main chat window without switching tabs. For developers who have already made ChatGPT part of their workflow, this is a significant simplification: instead of two separate tools, there will be a unified workspace. Codex will be available in the same dialogue as web search, image generation, file analysis, and — soon — the voice mode Bidi 1.

Agents — the third front

The third component of the large-scale update is tools for AI agents. ChatGPT is systematically transforming from a reactive chatbot into a proactive assistant capable of executing multi-step tasks autonomously: booking tickets, comparing products on marketplaces, filling out online forms, searching and synthesizing information from different sources.

"We want to make

ChatGPT an app that does everything for you that you normally do on the internet," sources familiar with OpenAI's strategy say.

All three components — live voice Bidi 1, embedded Codex, and autonomous agents — reinforce each other. An agent with duplex voice and the ability to write code is a fundamentally different class of tool, not just an updated chatbot.

What this means

OpenAI is deliberately moving toward a super app format — a single product that covers most everyday scenarios: voice, code, task automation, information search. For the average user, the immediate consequence is that conversations with ChatGPT will become noticeably more lively and natural. For business, this means a new level of automation of operational tasks that previously required a human or a set of separate SaaS tools. Competitors — from Google to Anthropic — will have to accelerate again.