The ChatGPT desktop app puts three modes in one window: Chat for questions and writing, Work for office tasks with several steps, and Codex for changing software code. Use Chat by default, and switch only when the job needs more than a conversation.
Imagine you open the desktop app after a week of using ChatGPT in your browser. Where you expected one chat box, you see three modes, and you are not sure which one to use for a quick email. This post shows you which mode fits which job, so you do not turn a small task into a long, costly one.
This post continues the series that teaches ChatGPT from scratch. The earlier post covered setting up an account, asking a real question, sharing a safe file, and asking for a rewrite. Here you will learn what each mode is for, what the Free plan still limits, why some modes use up your allowance faster, and how to choose without overthinking it.
Model names inside the app change often, so go by whatever the app shows today instead of writing a specific name into your team’s instructions. For the matching map of Anthropic’s Claude products, see Learn Claude from scratch. All learning paths are on Learn.
Why desktop is a different lesson
On web and mobile, almost everyone lives in conversation: you type, you get a reply, you rewrite, you leave. The desktop app still does that. It also packages agent-style office work and coding work into the same install, so you stop juggling three separate apps every time OpenAI merges its product lines.
OpenAI’s mid-2026 desktop direction is explicit: the ChatGPT desktop app for Mac and Windows is meant to hold Chat, Work, and Codex together in one place. The older Codex app can update into this combined app, and older “ChatGPT Classic” desktop builds may still exist as a migration path while people move over. UI (the screens and buttons you click) labels will keep shifting, but the three job types are the durable part of the map.
If you only need email rewrites and short explanations, web is still enough on its own. Desktop pays off once you want a dedicated window, local-feeling work patterns such as files, screen context, and coding projects, and a clearer split between modes than a single browser tab can show you.
Install and land, without a rabbit hole
- Open the official download hub at chatgpt.com/download, or the desktop feature page that OpenAI links from its help articles.
- Install for Mac or Windows from that page, and prefer the official page over any third-party mirror you find.
- Sign in with the same account you used on the web earlier in this series.
- Look for the mode switcher, labeled Chat, Work, and Codex or something close to it. If your build only shows a subset, note what you have and keep reading, because web Chat still covers conversation either way.
- If you already used a standalone Codex app, update it following OpenAI’s migration help. Existing Codex chats and projects are meant to carry over once the product merges cleanly.
You do not need every permission grant on day one. Folder access, browser control, and connector grants belong to later tutorials. Today is orientation: figuring out which mode fits which job.
Three modes, three jobs
| Mode | What it is | Best for | Not the first stop for |
|---|---|---|---|
| Chat | Conversation: you lead turn by turn | Drafts, rewrites, Q&A, learning, one-file review | Long unsupervised multi-app office jobs |
| Work | Agentic office-style multi-step help | Delegated tasks that need several steps, tools, or deliverables you still approve | A single paragraph rewrite you could finish in Chat |
| Codex | Coding-focused agent surface | Repos, diffs, PRs, developer workflows | Stakeholder emails and pure prose research |
Chat: conversation
Chat is the mode you already practiced earlier in this series. You own the loop: prompt, read, rewrite, export. Use it whenever the work is language-heavy, judgment-heavy, or small enough that an agent would just add ceremony on top.
Examples that stay in Chat:
- Rewrite a project update from bullet notes.
- Explain a metric definition for non-analysts.
- Turn a messy brief into three outline options.
- Review one PDF you are allowed to share and list the open questions.
Work: agentic multi-step office work
Work is the mode for doing more than one turn of typing on non-coding knowledge work. Think multiple steps: gather context, draft a longer artifact, use whatever tools you allow, and produce something closer to a finished package that you still review at the end. It is the product shape people compare to Claude’s Cowork-style agent work, even though the interface labels and permissions differ between the two.
Examples that belong in Work, when your plan and permissions allow it:
- Turn a folder of meeting notes into a structured brief with open questions attached.
- Build a multi-section report from several sources you explicitly provide.
- Run a longer delegated task where the intermediate steps matter and you want visible progress, not one shot of text.
Work is not a free pass to unsupervised action on company systems. Approvals, review, and data rules still sit with you the whole time. Later parts of the ChatGPT Work tutorial track go deeper on first agentic tasks; here you only need to recognize the mode and avoid wasting it on a one-line subject rewrite.
Codex: coding
Codex is the coding agent experience living inside the same desktop shell. It is built for developers and technical professionals: repositories, diffs, pull request review side panels, multi-repo projects, and computer-use style flows wherever the product supports them. If your job is SQL (the standard language for asking a database questions) drafts, Python scripts, or application code, Codex is the right door. If your job is a stakeholder email, it is the wrong door, with a higher chance of an expensive side quest along the way.
OpenAI’s public notes around the desktop merge mention coding workflow upgrades, such as inline edits in diffs, PR review in a side panel, and multi-repository projects, plus faster computer use powered by a current-generation model line marketed alongside the GPT-5.6 name. Hold that loosely, because your build may show a different model string next month. Read the name in your own interface rather than hardcoding last week’s blog post into your team wiki.
Plan reality: Free vs paid on desktop
Marketing lines often say Chat, Work, and Codex are available on every plan, Free included, on the desktop app. That can be true as an “you can open the doors” claim and still false as a “you get the same capacity behind them” claim. Free is useful for orientation, but multi-step Work and Codex tend to burn usage faster and hit limits sooner. Plus, Pro, Business, and Enterprise are where heavier Work and coding patterns usually become practical for everyday use.
Mobile notes matter for expectations too: download pages have described ChatGPT Work on phones with paid-plan language in places. Do not assume every desktop button has a full twin on Free mobile. When money or a service-level commitment is actually on the line, recheck OpenAI’s current plan matrix for your account type before you rely on it.
| Question | Practical answer |
|---|---|
| Can Free open the desktop app? | Yes, downloads are for global Mac and Windows use with a normal ChatGPT login. |
| Can Free touch Work or Codex? | Often yes, as a limited preview of the modes; capacity is the real constraint, not curiosity. |
| When should I pay? | When Chat limits or Work/Codex capacity actually block your weekly work, not because a banner said so once. |
| Business / Enterprise? | Org admin, data controls, and workspace choice matter more than mode labels here. Follow your IT policy. |
Mode chooser: pick before you type
Ask yourself three questions out loud before you switch modes:
- Is the main artifact language, or code? If it is language-first, use Chat or Work. If it is code or repo-first, use Codex.
- Is one careful conversation enough? If yes, stay in Chat. If the job needs multi-step tool use and longer delegated progress, move to Work, or Codex for engineering work.
- Am I willing to review intermediate steps? If you just need a thirty-second subject line, stay in Chat. Agents earn their keep when review is genuinely part of the job, not when you only wanted a faster autocomplete.
Quick scenarios
| You need… | Start in | Why |
|---|---|---|
| A shorter email from notes | Chat | One loop; agent overhead is waste |
| A multi-source brief with steps and files | Work | Multi-step office task |
| Explore a repo and propose a patch | Codex | Coding surface |
| Learn what a feature flag means | Chat | Q&A, not a project |
| Review a PR and comment on diffs | Codex | Built for that workflow |
| Turn three meeting PDFs into a plan | Work (if allowed) | Multi-step + deliverable |
If you pick wrong, stop early. Copy the useful text out, open the right mode, and start clean rather than dragging a coding agent through a press-release rewrite it was never built for.
Work and Codex burn usage faster
Chat is usually the cheapest way to spend a usage limit: one prompt, one reply, and maybe a couple of follow-ups. Work and Codex spend more because multi-step agents take more tool calls, more context, and more “thinking” time along the way. That is not a flaw in the product. It is exactly why Free feels generous for chat demos and tight for agent demos.
Habits that protect you:
- Scope the job in one sentence before you start Work or Codex, because vague agent goals burn tokens (small chunks of text, about three-quarters of a word each) while the model wanders around exploring.
- Prefer Chat for the first draft of language, and only move to Work if the remaining work is genuinely multi-step.
- Do not open Codex for prose. You will get tooling noise and usage burn for a job Chat already handles.
- Watch the plan usage indicators when the product shows them. If you hit a wall mid-task, export what you already have instead of re-running the whole agent loop from scratch.
- On shared Business seats, a heavy Work or Codex session can be a real team cost, not a private hobby, so know your org’s norms before you run one.
Rule of thumb: if the task fits in a single screen of notes and one deliverable paragraph, stay in Chat. If the task needs a checklist of steps and tools, consider Work or Codex instead.
Worked example: same goal, three wrong and right doors
Goal: turn last week’s messy product notes into a one-page status for leadership, then list three open engineering questions.
Wrong: Codex first
You open Codex because it looks powerful. The product expects a repository (a project folder that keeps the full history of its files), so you spend fifteen minutes explaining that there is no code at all. Usage burns while you do it. You get back a partial markdown (a simple way to mark headings and bullet points with plain symbols) file that still needs the same rewrite Chat would have finished faster.
Wrong: Work for a two-paragraph email
You launch Work for a short status update. The agent asks for extra context, tools, and steps you did not want to provide. You only needed a draft and a human edit pass. Chat would have finished in two turns with far less burn.
Right: Chat first, Work only if the job grows
Role: Draft for human review.
Audience: leadership + eng lead.
Goal: one-page status + 3 open engineering questions.
Tone: plain, no hype.
Constraints: 350-500 words. No invented metrics.
If a number is missing, say "unknown" instead of guessing.
Source notes:
[paste bullets]
Task:
1) Status draft with sections: shipped, in progress, blocked, next.
2) Three open eng questions (only from notes).
3) Assumptions list.
4) Unknowns list.If, after that draft, you also need to merge three PDFs, pull calendar items, and produce a formatted package with visible intermediate steps, then consider Work on a plan that allows it. The Chat draft still travels with you as the language spine of the final package.
Migration notes without the panic
People run into three confusing states on 2026 desktop installs:
- Old ChatGPT desktop (“Classic”) is still installed while the new combined app sits separately. Follow OpenAI’s migration prompt rather than running both forever.
- Codex app updates into ChatGPT desktop. That is intentional product-merge behavior, not a sign you installed the wrong thing.
- Mode labels lag the blog post. Community threads sometimes report a missing Chat button, or only Work and Codex on a build while web still shows Chat. Use web Chat if conversation is missing, and keep the three-job map in your head even when a button is temporarily absent.
None of that changes the teaching point: conversation, multi-step office agent work, and coding are three different jobs. The interface is still catching up to that split.
What you are not doing in this part
- Full Work approval loops and enterprise connector setup, saved for later Work tutorial parts.
- Repo hygiene, tests, and PR review discipline, saved for later Codex tutorial parts.
- Custom GPTs as a product, which belong to a separate series.
- API (a way for one program to ask another for data) keys and builder tools, covered in a later product map.
If you want the parallel story on Anthropic’s side, meaning chat versus Code versus Cowork, keep Learn Claude and the Claude product map nearby. It is the same judgment muscle, just different labels and install paths.
Common mistakes with the three modes
| Mistake | What happens | Fix |
|---|---|---|
| Treat every mode as “better chat” | Wrong tool, wasted usage | Use the three-job table |
| Start Work for a one-line rewrite | Overhead and burn | Stay in Chat |
| Start Codex for leadership prose | Repo-shaped friction | Chat (or Work if multi-step office) |
| Ignore Free limits | Mid-task walls feel like outages | Scope agent tasks; upgrade only when weekly work needs it |
| Grant every permission on day one | Risk without teaching value | Chat first; add grants when a real job needs them |
| Hardcode a model name in team docs | Docs rot when the interface renames it | Describe the job; note “current model string in UI” |
This week: try all three modes
- Install or open the official desktop app and list the modes you actually see.
- Run one pure Chat task, a rewrite or an explanation, and export it the same way you did earlier in this series.
- If your plan allows it, open Work once with a tiny multi-step job, such as “turn these three bullets into an outline and checklist, then stop.” Cancel it if it starts wandering.
- If you write code for work, open Codex once against a safe toy repo or a throwaway folder. Do not point it at production secrets on day one.
- Write yourself a sticky note: “Chat for language. Work for multi-step office. Codex for code.” Keep it up until the habit sticks on its own.
Questions people ask about this split
Do I need desktop if web Chat is enough?
Not really. Desktop stays optional until Work or Codex becomes part of your weekly work, or you simply prefer a dedicated window on your screen. The web and mobile habits from earlier in this series still count.
Is Work the same as Chat with a different skin?
They are not the same. Chat is turn-by-turn conversation that you drive yourself, while Work is built for multi-step agentic office tasks where you still review the outcome. The cost and the failure modes differ between them too.
Is Codex only for professional software engineers?
It is aimed at coding workflows, but analysts who live in scripts and repos can use it too. People who only need prose should not force every task into Codex just because the mode happens to exist.
Why do people mention GPT-5.6 with Work and Codex?
OpenAI’s public product notes around the desktop merge and its coding features have referenced that model family for some capabilities, such as computer-use notes. Names change often: by October 2026, OpenAI’s help page on managing usage in Work and Codex describes a newer model, GPT-6 Astra. Use the model label shown in your own interface, and OpenAI’s current docs, whenever accuracy matters for a team standard.
Where is the Claude version of this map?
Learn Claude from scratch and the Claude product map series cover chat, Code, and Cowork-style work on Anthropic’s side, teaching the same “pick the door for the job” lesson.
Quick recap
- Desktop ChatGPT for Mac and Windows is where Chat, Work, and Codex live together, downloaded from official OpenAI pages.
- Chat means conversation. Work means multi-step office agent work. Codex means coding.
- Free can open more doors than it can fully fuel, so usage burns faster on Work and Codex.
- Pick the mode based on the job, not on which button looks newest.
- Hold model names loosely; trust the job map and whatever the interface shows you today.
Series notes
This is Part 4 of Learn ChatGPT from scratch. The next post covers memory, custom instructions, and Projects, the sticky context that keeps you from re-pasting the same house rules every morning. For Claude’s parallel track, see Learn Claude. More paths: Learn hub.
Sources
Desktop packaging and plan limits change often. Re-check the official pages before you set a team standard on them.
- OpenAI: ChatGPT for your most ambitious work (desktop app with Chat, Work, and Codex; Free availability and Codex app merge)
- OpenAI Help: Moving to the new ChatGPT desktop app (Classic vs new app; Codex migration notes)
- Download ChatGPT (desktop and mobile download hub)
- ChatGPT Learn: What’s new (desktop Codex merge notes and related feature blurbs)
- ChatGPT web (conversation baseline when desktop modes lag or differ)
- OpenAI Help Center (plans, usage, account help)
- Analytics Made Simple: Learn Claude from scratch (sibling product path)
- Analytics Made Simple: Practical AI (vendor-neutral verification habits)
- Analytics Made Simple: Learn (learning hub)
Keep going
Same lessons in your feed
Short diagrams, hooks, and weekly tutorials on Substack, Instagram, X, and Facebook.
