---
title: "Two Tabs, Then One"
date: 2026-11-05
category: Companies & Stack
site: NeuroAI
canonical: https://neuroai.site/a/na-chatgpt-merge-25-1105
language: en
---

# Two Tabs, Then One

> On November 2, some paying ChatGPT users opened the desktop app and found an email waiting. It said the Chat and Work tabs would become one. It did not say when everyone else would get it, or what it

On November 2, some paying ChatGPT users opened the desktop app and found an email waiting. It said the Chat and Work tabs would become one. It did not say when everyone else would get it, or what it would look like.

The division of labor used to be easy to explain. Chat was for questions, one line in and a paragraph out. Work was for a job that takes a while: research, a slide deck, a website.

Separate tabs kept things clean. That was also the problem. Many people got halfway through a task before realizing they had started in the wrong place, and had to repeat what they had already said.

## The Email Arrives First

The wording of the email was careful. OpenAI said it would try the combined view with a small group of users over the coming weeks, gather feedback, and widen access later. It gave no date for a full rollout.

So November 2 was not the day a product launched. It was the day a notice arrived. If you did not get the email, the change is still real; you are simply not in the first group.

Gray release is a common practice in software. A small group steps on the landmines first, and then the company decides whether to let everyone else walk. For users, the downside is that the tutorial you find may not match the screen you see.

It is also what product teams do when features pile up: they rearrange them. The layout changes, the capabilities do not, but the path gets shorter and there is one less thing to remember.

## One Box, From Idea to Delivery

After the merge, users no longer have to sort a request in their head before typing: is this small talk or a piece of work? You can state an idea in one conversation and let it carry the idea forward without switching windows.

According to OpenAI, you can bring files and plugins into the process and ask it to build a website, make a presentation, or handle tasks on the web. You can watch it work and redirect it midway.

That sounds like a feature list, but it describes something else. You used to need a different entrance to work a different way. Now the entrance is gone, the way of working remains, stacked in one place.

For a writer, it means research can turn straight into layout. For a small business owner, it means a described requirement can turn into a page that opens. The step in between has been erased.

## The Coding Half Stays Outside

One exception was written into the email: Codex remains a separate part of the desktop app. It will not be pulled into the unified chat window, and it will not disappear with the tabs.

The arrangement is not hard to understand. Code has many files, long steps, and rollbacks when something breaks; the interface needs a different density. Stuffing it into a chat box would force two rhythms onto one desk, and neither would have room.

So the picture looks like this: everyday tasks move toward the center, professional workflows stay at the edge. The two major vendors made the same choice, though one opened its door wider than the other.

## The Bill Closes in December

The change that matters more than the interface is money. Starting in December, the Plus, Pro, and Business plans will share one allowance across ChatGPT and Codex, measured in tokens. The merge saves a click; the shared allowance changes how you use the product, because two budgets become one account.

This is neither a price increase nor a price cut. What changes is the unit of billing: from per product to per pool. OpenAI has not said how big the pool is, so users can only infer it from their own usage and learn the cost when the bill arrives.

A token is the unit a model uses to read and write text. The more complex the task and the stronger the model, the more of the allowance a conversation consumes. You do not need to memorize this; the Usage page in Settings shows the remaining allowance and the reset time.

One item was called out separately: Instant chat does not draw on the shared allowance. The lightest kind of question-and-answer stays outside the account, so a quick question or a reworded sentence will not move the water line.

OpenAI says most users will not need to wait for a reset more often than they do now. It also left room: depending on usage and the model chosen, some heavy users may hit the ceiling more often.

In plain terms: if you only chat, you will barely notice. If you write code in Codex by day and run long tasks in Chat at night, both taps draw from the same pipe, and either one lowers the level.

When the allowance runs out, the page offers three options: wait for a reset, upgrade the plan, or buy credits for eligible features. Three sentences that put the cost of choosing on the table and take pricing back into the company's hands.

## A Rival Got There Six Weeks Earlier

This is not one company's move. On September 16, Anthropic had already merged Claude's chat and Cowork into a single interface, ending manual switching and letting the model decide which capability to call. The user only has to say what they want.

In the same update, Claude also introduced documents and slides, first for Pro and Max users. Merging the entrance and completing the output are two steps of one idea: keep the person, help them finish the job, and let them take the result away.

From September 16 to November 2 is about six weeks. Same direction, one after the other. The first mover had already put it in front of paying users; the second was still testing with a small group.

Two companies arrived at the same shape, which suggests the direction is clear. Once entrances pile up past a certain point, users stop wanting to learn them. Rather than teach people how to choose, remove the choice.

The boundaries they kept are similar too: merge the general, separate the professional coding tool. Codex stays outside, and so does Claude Code. That is not a coincidence; it is the same question answered twice, with nearly the same steps.

## Fewer Doors, a Clearer Bill

For ordinary users, the most direct change is one less click. A click is cheap. What is valuable is no longer having to decide what kind of task this is, and no longer having to repeat yourself because you picked wrong.

For paying users, the real adjustment is different: two pockets have become one. The extra rounds you run in Codex come out of the same pocket as chat, so the bottom arrives sooner.

That is why an interface change has drawn so much discussion. It moves more than the position of a button. It moves the boundary of what you are actually buying, and the feeling of watching that boundary be redrawn.

A chat box and a work box, merged into one. Fewer boxes, a clearer bill, fewer choices. Whether it is easier to use will be settled by the people who received the email first.

**Disclaimer: This article is based on publicly available information and is for reference only. It does not constitute investment advice.**

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-chatgpt-merge-25-1105
Free to quote with attribution and a link to the original.
