ChatGPT Plus vs Pro (US$100 · 5x) vs Pro (US$200 · 20x): Price, Usage and How to Choose in Hong Kong

Choosing between ChatGPT Plus, Pro (US$100 · 5x) and Pro (US$200 · 20x) looks like a choice between three prices. It is really a choice between three intensities of work. Read only the labels "5x" and "20x" and you will mistake a relative usage level for a message bundle or an unlimited plan. Read only the monthly fee and you will ignore what hitting a limit actually costs you in blocked hours.

Facts checked: 15 September 2026. This guide is built from the OpenAI plans and Codex pricing page, the ChatGPT pricing page, the Plus and Pro help articles and the model-usage documentation as they stood that day. Every official price below is a US price. Plans and limits change often, so re-open the official page before you pay.

New sign-ups and upgrades to the US$200 tier are paused. OpenAI's Pro tiers page states that "As of September 10, 2026, we're temporarily pausing new sign-ups and upgrades to the ChatGPT Pro $200 plan". Existing US$200 subscriptions and all Pro US$100 subscriptions are unaffected. The public pricing page now shows Pro as "From $100 / month".

Supported regions (checked 15 September 2026): Hong Kong does not appear on OpenAI's published list of supported countries and territories for ChatGPT, nor on the API supported-countries list. The ChatGPT page states that "Accessing or offering access to our services outside of the countries and territories listed below may result in your account being blocked or suspended." This comparison is for plan research only. It does not mean you are eligible to open an account or to pay, and it does not offer a VPN, an overseas address, a borrowed phone number or anyone else's identity as a workaround. Read the pre-subscription checklist for Hong Kong readers (Traditional Chinese) first.

Every plan at a glance

OpenAI's plans and Codex pricing page also states that "GPT-5.3-Codex-Spark is in research preview for ChatGPT Pro users only", which now covers both the US$100 and US$200 price points. If you are weighing Go against Plus rather than Plus against Pro, the Go vs Plus comparison (Traditional Chinese) covers that split in detail.

The USD column is OpenAI's US list price. The HK$ column is our all-in service price on the day of writing. They are not the same product converted at today's exchange rate: who you pay, what is delivered, how refunds work and who carries account responsibility all differ. Our HK$ prices are also not OpenAI's official Hong Kong prices, because OpenAI does not publish Hong Kong as a supported market.

What "5x" and "20x" actually mean

OpenAI's page on the Pro tiers words it precisely: the US$100 tier is "5x higher usage than Plus" and the US$200 tier is "20x usage than Plus". Both numbers are multiples of the Plus baseline. None of the following follows from them:

• a fixed number of messages you may send each month;
• a bundle of tokens you can allocate however you like;
• every model and feature draining the allowance at the same rate;
• immunity from fair-use, capacity or safety limits;
• permission to split the account between five or twenty people.

How much work you actually finish depends on the model, context length, files you upload, image generation, research depth, the size of the Codex repo, tool calls and output length. Specific models are also documented separately: OpenAI's page on GPT-5.6 and GPT-6 Pro in ChatGPT publishes its own per-model availability and allowances per plan, which are separate from the 5x and 20x multipliers, and it warns that "model availability and limits change over time, check the model picker for the options on your account". The most reliable source is therefore what your own account shows today plus your own usage log.

The same distinction is worth keeping in mind if you are also pricing Anthropic: Claude's Max tiers use their own baseline and their own session mechanics, and the two vendors' multipliers are not interchangeable. See the Claude Pro vs Max comparison.

Plus: the sensible starting point for most individuals

The official ChatGPT Plus page lists what the paid personal plan includes, while noting that "Model availability changes during rollouts; use the model picker for current access" and that Plus "may include usage limits such as message caps, especially during high demand". Plus fits when:

• you write, translate, analyse documents or organise information a few times a week;
• you use images, research or Codex occasionally rather than all day;
• you are upgrading from Free for the first time and need to prove the paid features actually save you time;
• hitting a limit means switching to other work, not losing a delivery.

The real value of Plus is that it gives you a baseline. Spending a month watching which tasks pay off and which features burn the allowance fastest is almost always more sensible than buying the top tier because you are afraid of running short.

Pro (US$100 · 5x): buying headroom for work that keeps stalling

The test for Pro (US$100 · 5x) is not "I use it a lot". It is "Plus limits repeatedly stop work that matters". Signals worth acting on:

• you hit a limit several times a week, in the middle of focused work;
• your research, coding, document or image work needs long context held open;
• waiting for a reset genuinely delays a client deliverable, a release or an operational task;
• you have already tried splitting tasks, deleting irrelevant context and using lighter options, and it is still not enough.

If your heavy usage lands on only one or two days a month, read OpenAI's page on using credits for flexible usage in ChatGPT before upgrading. Plus and Pro subscribers can buy credits for supported features, credits "are valid for 12 months from purchase" and unused ones expire without rolling over, and they "currently can only be used with Codex, ChatGPT Work, and ChatGPT for Excel, when those features are available to your plan". Set a budget first, and do not assume topping up is always cheaper than upgrading.

Pro (US$200 · 20x): only when 5x is provably not enough

The US$200 tier suits people for whom ChatGPT or Codex is already a primary production tool: several large repos a day, long research, heavy image work or parallel projects. Before you consider it you should have two pieces of evidence:

1. Pro (US$100 · 5x) still runs out under a normal, already-optimised way of working;
2. the monthly value of time lost to waiting, switching tools and rebuilding context exceeds the upgrade difference.

Without those two records, 20x easily becomes expensive peace of mind. A bigger allowance also does not mean better output: weak prompts, bloated context and no acceptance criteria simply burn it faster.

There is also a practical constraint. Since 10 September 2026 the US$200 plan can only be kept, not newly started or upgraded into. If a service claims it can open a fresh US$200 subscription for you, ask how it handles the pause and whether it involves a shared or transferred account. OpenAI's own Pro tiers page points to terms that prohibit "Sharing your account credentials or making your account available to anyone else" and "Reselling access or using ChatGPT to power third-party services", so no arrangement built on shared credentials should be described as safe. Our own Pro (US$200 · 20x) service is likewise limited to renewing an existing subscription for now.

Pick by workload, not by price tag

Use a seven-day log, not a usage chart you found online

Record five things a day: the task, the model or feature, start and end time, whether you hit a limit, and how many minutes you were blocked. After seven days, do the arithmetic:

estimated monthly cost of being blocked = blocked hours per week x 4.3 x what an hour of your time is worth
net value of upgrading = estimated monthly cost of being blocked - the real price difference

The point is not to monetise every minute. It is to stop deciding because "5x sounds like a lot". The full logging method, and what to do once your Codex allowance runs out, are in the ChatGPT and Codex usage limits guide.

Six things to re-check before you pay

1. whether the official supported-countries list and the terms of use allow your actual situation;
2. whether you are buying an official self-service subscription or a third-party service, and who receives the payment;
3. for any managed or pre-made account, who controls the login, the email address and the recovery details;
4. whether it auto-renews, when the cancellation cut-off is, and who handles refunds;
5. whether the figure quoted is OpenAI's list price, the amount your card issuer will settle, or a merchant's all-in service price;
6. what the merchant can and explicitly cannot guarantee if the account is flagged for verification, a payment is disputed, or access is suspended.

Use the HKD cost calculator to compare exchange rates against an all-in service price, then the AI plan comparison tool to filter by workload. If you understand the eligibility, account and delivery risks above and want to see what we offer, the ChatGPT service category lists the individual pages for Plus, Pro (US$100 · 5x) and Pro (US$200 · 20x), currently renewal-only. HK Learn AI is an independent merchant and is not affiliated with OpenAI or Anthropic; the HK$ amounts on those pages are our own service prices, not OpenAI's official Hong Kong prices, how it works explains the delivery model, and what is delivered is governed by our terms of service.

Bottom line: start at the lowest plan that is enough

Plus is for building a baseline. Pro (US$100 · 5x) is for people who have already shown that Plus limits obstruct professional work. Pro (US$200 · 20x) is for the heaviest users for whom 5x still falls short, and new sign-ups and upgrades to it are paused for now. Neither 5x nor 20x is a fixed quota or a promise of unlimited use.

Clear the region and account eligibility question first. Then let a seven-day log answer the only question that matters: how much time do these limits actually cost me? That answer will choose your plan better than any online estimate or straight price comparison.