Free, Go, Plus, Pro, Business, Enterprise without hype

Pick a ChatGPT plan by how you actually work: how often you use it, what kind of information you put in, and who should pay. Start there, not with the price page.

Imagine a red banner says you have hit your limit, and you upgrade to the most expensive personal plan that same week. Or you stay on Free for months and decide AI chat is a toy. Or your team buys personal plans on their own cards and pastes customer lists into them. Each one is a reaction to a price page, not a decision.

This post compares Free, Go, Plus, Pro, Business, and Enterprise the way you would compare phone plans. Prices change and can differ by country, so check the ChatGPT pricing page (checked September 2026) before you pay. For the same comparison for Claude, see Learn Claude from scratch.

The plan ladder in plain English

OpenAI’s personal and business plans currently look roughly like the list below. Tax, currency, regional availability and promotional wording all vary, so confirm on the live page. Older guides that list only Free, Plus and Pro are incomplete, because Go is now a real middle tier.

• Free: $0, as of September 2026 on the ChatGPT pricing page. It is a real product with real limits, and it is fine for light exploration and occasional drafts.
• Go: a low-cost personal subscription, often shown around $8 per month as of September 2026 on the ChatGPT pricing page. It gives you more messages, uploads, image creation and memory than Free. OpenAI’s help pages say this plan may include ads and that some advanced products still need Plus or higher, so confirm the price and the ad policy before you subscribe.
• Plus: often shown around $20 per month as of September 2026 on the ChatGPT pricing page. This is the usual rung for a serious individual. It adds advanced models, Projects, scheduled tasks and custom GPTs (saved ChatGPT helpers set up for one job, GPT being OpenAI’s name for its AI models). It also gives you more room for the Work and Codex features (explained below) and earlier access to new features. Usage is still finite, even when marketing says “unlimited*” and the fine print mentions abuse limits.
• Pro: starts higher, with two tiers that OpenAI has framed around $100 and $200 per month, as of September 2026 on the ChatGPT Pro help page. Pro includes everything Plus does and much more usage. That help page describes roughly 5 times the Plus usage at the lower price and roughly 20 times the Plus usage at the higher one, with the same core capabilities. The strongest reasoning models and the largest allowances for Codex, research and images live here.
• Business: team seats with an admin panel, one central bill, workspace controls, and privacy terms under which business data is not used to train models by default. It is priced per user per month, annual and monthly options often differ, and a minimum user count applies (commonly 2).
• Enterprise and Edu: custom contracts arranged with a sales team, with deeper admin controls. Those include single sign-on (SSO, one company login for every tool) and automatic adding and removing of user accounts. They also include data retention settings, choices about where data is stored, and compliance features. Education offers exist where available. This is not a personal upgrade button.

Treat those multipliers as orientation, not a promise of an exact number of messages. Long chats, big files, heavier models, Work runs, Codex sessions, deep research and image generation use up your allowance faster than short questions do. OpenAI’s own notes mention abuse limits and temporary restrictions when demand is high, so if a page says “unlimited*,” read what the asterisk says.

How usage limits actually feel

People often read a limit as “ChatGPT got dumber.” Usually the model is fine and your allowance is used up. Sometimes one model’s share is gone while others still work.

Capacity is not the same as quality

Upgrading can give you better models and more room, but it cannot turn a vague prompt into a good decision. A crisp session on Free with verified numbers still beats a sloppy Pro session that invents a metric. Payment buys capacity and features, and it does not buy truth.

Different tools use up the allowance differently

A three-line email rewrite costs almost nothing. A multi-hour Codex session costs far more, and so does a deep research run, a large file analysis or a long Work deliverable. Codex is OpenAI’s coding assistant, and Work is the feature that carries out multi-step tasks for you. If you use browser chat, mobile voice and desktop Work on the same plan, they all draw from one account’s limits. That surprise causes many Pro upgrades that could have been solved by not running three heavy jobs at once.

Model-specific allowances on Pro

OpenAI’s help article on Pro says some models have their own separate allowances, and the lower-priced Pro tier gets smaller ones than the higher-priced tier. If you hit a model’s allowance, that model may be unavailable until it resets, and your subscription keeps running. Check the reset time in the product screen before you assume the whole account is broken.

What each step up promises compared with Free

Free is intentionally limited in messages, uploads, images, memory, deep research, Codex and Work. Work on Free is often limited to the desktop app. Go expands the popular everyday tools. Plus adds advanced models and productivity features. Pro multiplies the room and adds the strongest reasoning models. Use these steps as a ladder and not as a ranking of how serious a professional you are.

Free: what you actually get

Free is not a demo video. You can chat on the web, iOS and Android, and you can write and edit. The pricing comparison chart also lists limited image generation, deep research, memory, Codex access and desktop Work. The lasting point is simpler: Free is capacity-constrained, and you hit the ceiling when you try to make ChatGPT your all-day coworker.

Free fits you if:

• You draft the occasional email, outline or explanation.
• You are testing whether the product suits how you think before paying for it.
• You already have another paid assistant and want ChatGPT only as a second opinion.
• Your employer has not approved paid consumer AI spending yet, and you stick to practice tasks that contain nothing sensitive.

Free will frustrate you if:

• You hit limits in the middle of a project several days a week.
• You need advanced reasoning models, fuller Work on web and mobile, or serious Codex as a daily tool.
• You rely on larger context (how much text the model can keep in mind at once), more uploads or scheduled tasks that the current chart does not give Free.

Frustration is useful information. If Free blocks real work, upgrade for the capacity. If Free is fine and you are upgrading because a podcast host said Pro is “mandatory,” that is fear of missing out and not a reason.

Go: the middle tier for people who outgrew Free

Comparing ChatGPT Go vs Free comes down to message volume and file limits. While the Free tier offers baseline chat access with hourly caps, Go offers more chat volume, extra image generation, and higher upload limits at an affordable price. It fits everyday users who outgrow Free limits but do not need heavy developer tools.

Important caveats from OpenAI’s help pages:

• Some advanced products and higher-tier features still require Plus or another eligible plan.
• The plan may include ads, and OpenAI has published help material about testing ads on Go. If ads bother you more than the price gap to Plus, count that preference honestly.
• API usage is not included and is billed separately. An API is a way for one program to ask another program for data or an action.
• Legacy (older) model access is not promised on Go the way it can be on higher plans.
• Individual Go, Plus and Pro subscriptions have been monthly and not annual in the current help FAQs. That can change, so check the billing documents if it matters to you.

Go is a sensible step if Free limits interrupt useful work most weeks, you mostly need more chat and file room, and you do not yet need the full Plus bundle. Go is optional if you open ChatGPT twice a month, or if you already know you need advanced reasoning, the full Work features or the early access that Plus advertises.

Plus: the usual paid plan for one person

Plus is where OpenAI expects many everyday buyers to land. The headline bundle is everything in Go. On top of that come advanced reasoning models, more messages and uploads, stronger image creation and expanded deep research. You also get more memory, Projects, scheduled tasks, custom GPTs, more Codex use, Work on desktop, web and mobile, and early access to new features.

More room and better models

This is the benefit you feel first. Sessions last longer before a limit message appears. You also get advanced reasoning models that Free and Go may not fully include, and there is room for drafting in rounds: rewrite, rewrite again, then ask for a skeptical pass. If your pain on Free or Go was a banner that killed your train of thought, or a need for the smarter model on hard tasks, Plus is aimed at you.

Work and Codex access

Plus widens Work beyond the desktop-only version on Free and Go. It also raises Codex usage compared with the lower tiers. That matters if multi-step deliverables or coding are real parts of your job. Heavy use still eats into your capacity, so developers who work in code all day often outgrow Plus and look at Pro or pay for API credits.

Projects, tasks and custom GPTs

Projects keep chats and documents organized for ongoing work, such as a client, a course or a system migration. Scheduled tasks help when the product supports recurring or trigger-based jobs. Custom GPTs are reusable assistants that carry your instructions and tools. Go may include some of these with limits, and Plus is usually where they feel like a daily workbench, though exact availability still depends on the live comparison chart.

Plus is a sensible default if you use ChatGPT several times a week for real deliverables and want advanced models. It also fits if you want expanded Work and Codex and hit Free or Go limits often enough to notice. Plus is optional if you open ChatGPT twice a month for fun outlines, or if Go already covers your real volume.

Pro: when more capacity is the product

OpenAI offers two capacity tiers within ChatGPT Pro. The tier gives roughly 5x the usage allowance of Plus, while the tier gives about 20x the capacity of Plus. Both tiers share the same core models, but expand allowances for heavy coding, deep research, and long automated agent workflows. An agent is a program that can edit files and run commands on its own.

• Pro (lower tier, often about $100 a month): OpenAI’s help describes about 5 times the usage of Plus, the same core Pro capabilities as the higher tier, and smaller model-specific allowances than the $200 tier.
• Pro (higher tier, often about $200 a month): the same core Pro capabilities with about 20 times the usage of Plus. This is power-user territory, such as long coding days, constant agent-style runs, and research-heavy roles where waiting on a limit banner costs more than the subscription.

Buy Pro when you can point to a pattern, such as “I hit Plus limits on most workdays while doing useful work I know how to check.” Do not buy it because a stranger online said the top tier is the only serious one. Plenty of excellent work happens on Plus, and plenty of mediocre work happens on Pro with unchecked SQL and confident wrong numbers.

If you are a developer comparing Pro to paying for the API directly, run the math for your token volume (a token is a small chunk of text, roughly a word). API prices are public in OpenAI’s platform documents, while consumer Pro is a subscription with shared limits and abuse rules, and neither is cheaper for everyone.

A comparison table you can actually use

These values are orientation only. Confirm them on the live ChatGPT pricing page and in OpenAI’s help pages, because the model family names (the GPT-5 labels) will change.

The API platform is a separate wallet

Two OpenAI logins confuse people for months, so here is the split.

• ChatGPT subscription (Free, Go, Plus, Pro, Business, Enterprise): for people chatting with the product and the tools that come with it, metered by plan limits and abuse rules.
• OpenAI API and developer platform: keys for applications, automation and builders. You pay per use for models and related tools, and the price tables live in the platform documents, not only on the ChatGPT consumer pricing page.

You can hold both, and many developers do, using Plus or Pro for interactive chat, Work and Codex and API keys for apps. Paying for Pro does not automatically fund unlimited API traffic. An API budget does not automatically include every consumer chat perk either, and the Go help page says explicitly that API usage is billed separately.

If your job is talking to ChatGPT in a window, you can ignore the API until a later post. If your job is shipping a feature that calls a model, start with the platform documents and your finance team’s cloud purchasing process, not only a personal Pro card.

Plus versus business: key differences between plans

The primary difference between ChatGPT Plus vs Business comes down to team management and data privacy. Plus is a personal account designed for a single person. In contrast, Business is a collaborative workspace built for two or more users with central billing, team admin controls, and default privacy protections.

On a personal Plus account, data privacy defaults to allowing model training unless you turn it off in settings. In contrast, ChatGPT Business excludes company chats from training by default. Business also gives team members a shared space to build custom GPTs together without expensing separate personal cards.

Business and Enterprise: when the company should buy

Personal plans are for individuals. Once several people need shared admin, one company login, central billing or clear rules about training on their data, look at the company plans.

OpenAI’s pricing and business pages describe these options:

• Business: for businesses, typically starting at 2 users, with per-user monthly pricing and monthly or annual options. It includes workspace admin, unified billing, connectors and team controls, and privacy terms under which business data is not used to train models by default. Exact seat prices and Codex packaging change, so verify them on the live business pricing page.
• Enterprise: sales-assisted and built for scale. It has deeper admin and compliance options such as single sign-on, audit logs, data residency choices and enhanced support, all shown on the comparison tables. Pricing is custom.
• Edu and education offers: separate paths such as ChatGPT Edu for universities and time-limited teacher offers in some regions. The security and admin features differ from personal Free. Contact sales or the education product pages if that is your world.

Smell test: if five employees each buy Pro on personal cards and paste customer data into personal accounts, you do not have a productivity win. You have shadow IT, meaning tools that people use for work without the company knowing. Business or Enterprise plus training on the rules is the responsible move, even if the sticker price looks higher than five Plus seats.

Also remember that on consumer Free, Go, Plus and Pro, OpenAI’s pricing comparison says your content may be used to train models unless you use the opt-out. Many business and Enterprise privacy terms start from the opposite default, so read the current data controls for the plan you actually use.

A plan chooser that avoids fear of missing out

Walk through these questions in order and stop at the first plan that fits.

1. Is this work with company data? If yes, ask IT or your manager which ChatGPT path is approved, and do not invent a personal Pro workaround for regulated data.
2. How many days a week do you hit Free limits while doing useful work? If rarely, stay on Free. If most weeks, consider Go before leaping to Plus.
3. Do you need advanced reasoning models, expanded Work on web and mobile, serious Codex, deep research or early features as a regular tool? That points to Plus or a company plan and not to heroic efforts on Free. Go may still be enough if your only pain is the volume of simple chats.
4. On Plus, do you still hit limits on most full workdays? Track it for two weeks. If yes, Pro 5x is the usual next step, and if you use agents or code all day and 5x still runs out, Pro 20x becomes worth a look.
5. Are you building software that calls models in production? Budget the API (and any cloud partner options) separately, because a chat plan will not replace the cost controls that engineering needs.
6. Are several people involved? Price Business or Enterprise against the mess of personal cards, and include admin time and data risk in the comparison.
7. Do the ads on Go bother you more than the price gap to Plus? Your preference is a real input, and neither answer is a moral failing.

Notice what is missing from that list: which model is winning benchmarks this week. Benchmarks are interesting, but they rarely justify a $200 personal subscription if your real job is rewriting status updates twice a week.

Cost scenarios with rough math

These are teaching scenarios with round numbers and not invoices. Your taxes, currency, regional pricing and promotions will differ, so check the live page.

Scenario A: the curious analyst

You have three sessions a week, short summaries and no Codex. Free is enough. If a busy quarterly week feels cramped, a single month of Go or Plus can be a temporary boost. You can cancel in the billing settings by following the live instructions.

Scenario B: an individual contributor who writes and codes lightly

You chat daily, get light coding help weekly and keep a couple of Projects. Plus at about $20 a month is the common default. If you mostly need more room than Free and do not care about advanced models yet, try Go for a month and keep a diary of every time you hit a limit.

Scenario C: full-time coding with agents

You have Codex or Work open most of the day, work in large code repositories, and often use the higher-tier models. Pro 5x or 20x may be reasonable here, so compare it with API-heavy workflows and any seats your company provides. Keep verifying the output, because a higher plan does not reduce the need to check AI-written SQL or to review code changes.

Scenario D: a ten-person team with mixed roles

Ten personal Plus seats cost roughly $200 a month before anyone buys Pro, and they add the hidden data risk described above. Business seats may come to a similar monthly amount with single sign-on and admin, depending on discounts and packaging, so ask for a real quote. The point is how the company sets things up, more than the arithmetic.

Common plan mistakes

• Upgrading before learning to prompt. A vague one-liner wastes Pro as efficiently as it wastes Free, so practice on Free for a week with tasks that contain nothing sensitive.
• Skipping Go when Free is your only pain. If your only issue is message volume, Go may be enough, and jumping straight to Pro for a daily status rewrite is theater.
• Ignoring the shared pool. Chat, Work, Codex and mobile quick questions all draw from one meter on a personal plan, so space out heavy jobs when you are near a limit.
• Using personal Plus or Pro for secret work data. A higher plan does not make you compliant with company rules. Following your company’s policy does.
• Assuming annual billing is always available. OpenAI’s help pages have said Go, Plus and Pro are monthly only, while Business and Enterprise may offer annual options, so confirm before you plan a yearly budget line.
• Mixing API bills with chat bills in your head. Keep separate budgets, and separate owners when you can.
• Chasing model names instead of fixing your process. Heavier models help with hard tasks and also use up your limits faster, so match the model to the job.
• Treating the assistant as the official record because you pay more. Payment buys capacity and features, not truth, and responsibility stays with you. Company numbers should still come from the company’s own official data sources.
• Reading outdated guides that only mention Free, Plus and Pro. Go is part of the ladder now, and so are two Pro tiers, so re-check the official page the week you buy.

Practice: a two-week plan experiment

You do not need an elaborate spreadsheet for this. You need honest notes.

1. Stay on your current plan for ten workdays. Each day you use ChatGPT, jot down the minutes used and the main task type, such as writing, summarizing, coding or research. Note where you used it (web, mobile or desktop). Also note whether you hit a limit banner or a missing-feature wall.
2. Mark sensitive and non-sensitive tasks. If sensitive work showed up on a personal account, flag it for a policy cleanup no matter which plan you are on.
3. Count the real upgrade triggers. “I was annoyed once” is not a trigger. “Limit banners killed productive work on 6 of 10 days” is a trigger. “I need Work on the web for multi-step deliverables twice a week” is a feature trigger, which is different from a volume trigger.
4. Choose one step up at most. Go from Free to Go, or from Go to Plus, or from Plus to Pro 5x. Skip a rung only if you can name a concrete missing feature, not a vague wish to feel “serious.”
5. Open the official pricing page the day you buy. Confirm the price, the ad wording on Go, Pro 5x versus 20x, and whether Business is the real team answer. Write down the date so your future self knows when you checked.
6. Keep learning. Whatever you buy, schedule your first 30 minutes on web and mobile (the next post in this series) before you spend a month collecting chats you never reopen.

How this series talks about plans

For the rest of this series on learning ChatGPT from scratch, we will name Free, Go, Plus, Pro, Business or Enterprise whenever limits or features matter. We will not pretend every reader is on Pro, since many good habits work on Free with practice data that is not sensitive. We will also keep reminding you that screens and prices move, and that company policy beats personal convenience.

Checking answers and handling data safely for any vendor still lives in the series on practical AI. If you want the parallel Claude plan ladder for comparison shopping, use the series on learning Claude from scratch. This track stays on ChatGPT so the names stay stable in your head.

Quick recap

• The personal ladder is Free, then Go, then Plus, then Pro (5x or 20x). Business and Enterprise/Edu are company products and not personal status tiers.
• Go is real: more everyday capacity than Free, a lower price than Plus, possible ads, and still not every advanced feature.
• Plus is the common individual workbench. Pro is mostly extra capacity, plus the strongest models, for people who already use ChatGPT hard.
• Usage is shared across every way you reach ChatGPT, and heavy Work and Codex use up far more than short chats.
• API billing is a separate wallet from ChatGPT subscriptions.
• Team data on personal seats is a risk problem and not a way to save money.
• Paying more does not make ChatGPT the official source for company numbers.
• Re-check the live pricing page the week you buy, because prices and model names change.
• Next up is web and mobile in your first 30 minutes, so the plan you picked gets used well.

Sources

• ChatGPT plans and pricing (verify current numbers before you buy): https://chatgpt.com/pricing
• What is ChatGPT Go: https://help.openai.com/en/articles/11989085-what-is-chatgpt-go
• About ChatGPT Pro tiers (5x / 20x usage framing): https://help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers
• ChatGPT Work and Codex availability and modes: https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex
• OpenAI business pricing entry points: https://openai.com/business/pricing/
• How your data is used / model training opt-out context: https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance
• OpenAI Help Center billing collection: https://help.openai.com/en/collections/3943089-billing
• OpenAI API pricing (separate from ChatGPT subscriptions): https://openai.com/api/pricing
• Analytics Made Simple Learn Claude series (sibling plan ladder): https://analyticsmadesimple.com/series/learn-claude/
• Analytics Made Simple Practical AI series: https://analyticsmadesimple.com/series/practical-ai/
• Analytics Made Simple: how to check AI-written SQL: https://analyticsmadesimple.com/how-to-check-ai-written-sql/

Series notes

This is Part 2 of Learn ChatGPT from scratch. Part 1 covered what ChatGPT is, and the next post covers your first 30 minutes on web and mobile.

Merch on Etsy

Open store →

Git Version Control Shirt View on Etsy → Mean Squared Error Shirt View on Etsy → Naive Bayes Coffee Mug View on Etsy → Git Version Control Shirt View on Etsy →