OpenAI is about to change the way your team works inside ChatGPT — twice, in a few weeks. First, the separate Work tab is being folded into regular chat, so a conversation can turn into a finished deliverable without switching screens. Second, from December, ChatGPT and Codex will draw from one shared usage budget on Plus, Pro and Business plans. If you know how to use ChatGPT Work mode for small business tasks, the merge is good news; if you do not plan for the shared budget, it can stall your team mid-task, with a client waiting.
This guide explains what Work mode is, what is changing and when, what it costs, and a five-step setup.
Table of Contents
What just changed: Chat and Work become one workspace
On November 2, 2026, OpenAI emailed subscribers to say it is merging the Chat and Work tabs in the ChatGPT desktop app into a single experience. TechCrunch reported the email the same day. The rollout starts with a small group for feedback, then widens — no firm dates given.
The email’s framing is simple: “This will enable you to move from exploring an idea to having ChatGPT carry it through, all in one place. You can bring in your files and plugins and ask ChatGPT to build a website, create a presentation, or do things for you on the web, while reviewing and guiding its work.”
Two things stay separate: Codex “will remain a separate experience in the ChatGPT desktop app,” and the change so far covers the desktop app only — OpenAI has not confirmed whether web and mobile get the same layout. For a small business, the meaning is simple: the seam between asking a question and getting finished work disappears.
What ChatGPT Work mode actually does
Work is ChatGPT’s agent mode. OpenAI launched it in July 2026, replacing the older ChatGPT agent. Where Chat answers one message at a time, Work plans a task, breaks it into steps, uses tools, and keeps going until it has something finished to hand back. Unlike ChatGPT’s always-on agents, which run in the background, Work runs while you watch and review.
The test is simple. Chat is for questions and quick help. Work is for jobs with a clear outcome — a brief, a deck, an analysis, a recurring update or a file you can review and use. OpenAI’s own Work guide puts it that way: use Work when you want ChatGPT to complete a task, not just answer one.
Work can use your files, plugins and approved tools to retrieve information, create finished files and run workflows. On the desktop app it can also use local files, apps and the browser. Plugins connect it to tools you already use — Slack, Google Drive, SharePoint, email, calendars, CRM systems and project trackers — and you can name one by typing @ plus the app name.
Three features matter. Projects hold a job’s files, context and instructions, so every chat starts from the same background. Plan mode proposes a strategy first, and you approve it before anything real happens. And local versus cloud mode: local mode reads files off your disk, while cloud mode runs on OpenAI’s servers and keeps a task going with your machine off, but cannot touch local files.
The bigger change: one shared usage budget from December
The interface merge is the visible change. The budget change affects what you get for your money. From December 2026, the included usage allowance on Plus, Pro and Business plans will be shared across ChatGPT and Codex, measured in tokens. OpenAI’s Help Center usage-limits page says so, and the November 2 email confirms it: “Your use of ChatGPT and Codex will draw from the same allowance.”
OpenAI has not named an exact start date in December, nor published a token figure per plan or a conversion rate per model and task. That is the honest gap — you cannot budget precisely yet.
What is confirmed: instant chat does not count — OpenAI says “Instant chat will not draw from your usage.” Advanced models and complex tasks cost more. One place to check — remaining usage and reset times show under Settings, then Usage. And when it runs out, you wait, upgrade or buy credits.
OpenAI says most users will not notice a difference, while some may hit limits and wait for resets more often — a forecast, not a guarantee. One piece of good news: on October 29, OpenAI cancelled a planned Pro cut and doubled the Work and Codex allowances for Pro 100, 200 and 500 instead.
What it costs: the plans that include Work
Work is available on every plan in the desktop app, including Free with limited usage; paid plans add web and mobile access and much higher limits. Prices below are OpenAI’s US list prices, checked against its pricing pages in October and November 2026 — confirm them before you buy, as OpenAI has changed these tiers twice in two months.
| Plan | Price | Work and Codex allowance |
|---|---|---|
| Free | $0 | Limited usage on desktop only |
| Plus | $20/month | Full Work and Codex; estimated 15–160 messages per 5 hours on GPT-6.1 Sol, 5–45 on GPT-6 Astra, 350–3,000 on GPT-6 Luna |
| Pro 100 | $100/month | Larger allowance; currently no 5-hour limit |
| Pro 200 | $200/month | More usage than Pro 100; currently no 5-hour limit |
| Pro 500 | $500/month | Most included usage; only personal plan with Ultrafast mode |
| Business Standard | $25/user/month, or $20 billed annually (2-seat minimum) | Same 5-hour estimates as Plus |
| Business Premium | $125/user/month, or $100 billed annually | 5x Standard’s usage, no 5-hour limit |
Two warnings. The message ranges are OpenAI’s estimates, not fixed limits — a long job can use far more allowance per message than a short one. Pro and Premium are the only tiers without a five-hour window, so a team running agent jobs all day will feel the Plus cap first. For most small businesses, Plus at $20 a month or a Business Standard seat is the right start.
How to use ChatGPT Work mode for small business: a 5-step setup
Step 1: Install the desktop app and pick one pilot job
Work is strongest on the desktop app, because only there can it read local files directly. Install it where the work lives, sign in, and choose one recurring job to pilot — not five. A weekly report, a monthly client update or a supplier comparison all work: they repeat and end in a file.
Step 2: Set up a project for that job
Create a project for the pilot job and put its files in the primary folder: the template, last month’s finished version, the price list, the brand notes. Most people skip this step, and it saves the most time.
Step 3: Connect only the plugins the job needs
Open Customize, then Plugins, and connect the tools the pilot touches — Drive or SharePoint for files, Gmail or your calendar for dates, your CRM for customer details. Connect one tool at a time and test it. A plugin you never use is just another place your data can flow, so keep the list short.
Step 4: Write the brief like a work order, then approve the plan
Give Work a goal plus the files or apps it needs, in plain language: what the deliverable is, who it is for, what “done” looks like, and what it must not do. Turn on Plan mode for the first few runs and approve the strategy before it starts. A good brief: “Using the sales export here and last week’s report as the template, build this week’s one-page sales summary as a spreadsheet and a short email draft. Flag any product whose sales fell more than 10% week over week. Do not email anyone.”
Step 5: Review the output before it goes anywhere
Work hands back finished files, but finished is not the same as correct. Check the numbers against your source file and confirm names, prices and dates before anything reaches a customer. That review is the owner’s whole job — Work drafts, you judge. Never let it send an email, post a message or change a shared file without a human looking first.
The shared-budget trap: four rules for December
From December, your ChatGPT agent work and your Codex coding draw from the same pool. If your business uses both, the pool can drain twice as fast.
Route quick questions to Instant. Everyday questions do not touch the allowance — keep them in plain chat and save the budget for jobs that end in a file.
Batch agent jobs instead of sprinkling them. One long Work session uses the budget better than ten small ones that each re-read the same files.
Check Settings, then Usage, every Monday. A two-minute check tells you whether Friday’s big job is safe — or should wait for the reset.
Do not run heavy Work and heavy Codex in the same week on Plus. If a developer is running long Codex sessions, schedule the big Work jobs for another week, or move the heavier job to a Pro seat.
Work versus the alternatives
OpenAI is not alone: Anthropic merged Claude’s chat and Cowork in September, and Google’s Gemini Agent for Work offers the same idea inside Workspace.
What to do this week
The merge has no firm date and the shared budget lands sometime in December, so this week is preparation, not panic: install the desktop app and check whether your account already shows the merged layout; open Settings, then Usage, and screenshot your current limits; set up one project for one recurring job; run it through Work once with Plan mode on; and decide who gets the shared budget first in December.
Frequently Asked Questions
When does the shared ChatGPT and Codex usage allowance start?
In December 2026, on Plus, Pro and Business plans. OpenAI has not named an exact date. The allowance will be measured in tokens, and Instant chat will not draw from it.
Does merging Chat and Work change what Work can do?
No. The merge removes the tab switch: agent capabilities that lived under the Work tab become reachable from the main chat surface. Codex stays separate inside the desktop app.
Is ChatGPT Work available on the free plan?
Yes, on the desktop app, with limited usage. Paid plans add web and mobile access and much higher limits — Plus at $20 a month is the usual start.
What happens when my shared allowance runs out?
You wait for the reset, upgrade to a higher plan, or buy credits for eligible features. Your combined remaining usage and reset times show under Settings, then Usage.
Conclusion
ChatGPT Work mode is the part of ChatGPT that finishes jobs instead of just answering questions — and from December it shares one usage budget with Codex while its tab merges into chat. Set up one project, connect only the plugins the job needs, write briefs like work orders, approve the plan, and review every output before it reaches a customer. Check your usage every Monday, and December becomes a line in your routine instead of a surprise mid-deadline.
Sources
- TechCrunch, “OpenAI is finally merging ChatGPT’s Chat and Work tabs,” November 2, 2026 — https://techcrunch.com/2026/11/02/openai-is-finally-merging-chatgpts-chat-and-work-tabs/
- OpenAI, “Get started with ChatGPT Work” (learn.chatgpt.com) — https://learn.chatgpt.com/docs/get-started-with-work
- OpenAI Help Center, “ChatGPT and Codex usage limits” — shared token allowance from December 2026; Instant chat excluded; Settings → Usage (cited via TechCrunch and independent coverage; no direct URL quoted in sources)
- Justin McKelvey, “ChatGPT Pricing (2026): Every Plan, Free to $500 Pro” (checked October 19, 2026) — https://justinmckelvey.com/blog/chatgpt-pricing
- Justin McKelvey, “ChatGPT Usage Limits 2026: Plus, Pro & Business Explained” (updated November 1, 2026) — https://justinmckelvey.com/blog/chatgpt-usage-limits
Note: OpenAI has not published a per-plan token figure for the shared allowance or a conversion rate per model/task — the article states this gap explicitly rather than inventing numbers.
Was this article helpful?