Blog · ChatGPT · Published
GPT-5 and 4o: Which Can You Use? How Version Numbers Map to the Models Available in Your Account, and How to Verify Them
A GPT version number is a model name; it is not the same as the list your account can actually call. Which models you can use is decided jointly by your plan tier, whether the model has been retired, and the allowance available at the time: older models such as 4o are being retired, Go does not include older models, Think uses GPT-5.6 Luna, and only Pro 500 includes Ultrafast. Always verify against the model picker.
GPT-5 and 4o: Which Can You Use? How Version Numbers Map to the Models Available in Your Account, and How to Verify Them
Information updated 2026-11-19
A GPT version number is a name OpenAI gives to a particular model; it is not the same thing as the list of models you can actually call in your account. Deciding "can I use GPT-5, 4o and the rest" does not depend on memorising every name. It depends on three things: your plan tier, whether that model is still in service, and the allowance remaining on the account at that moment. There is only one reliable way to check — open the model picker in the ChatGPT interface (the drop-down menu next to the input box for switching models) and see which options are listed for your own account right now.
1. Why version numbers and "the models you can use" do not line up
Many people assume that "a new official version = everyone can use it". That is the first misconception. OpenAI states plainly in its ChatGPT Plus article that model availability and allowances change over time, and that you should check the options currently listed for your account in the model picker; availability can also change while a model is being released in stages, so the model picker is always the reference.
The second misconception is treating an older model as permanently selectable. The same article links to "Retiring GPT-4o and other ChatGPT models", which explains that older models such as 4o are in the process of being retired and that there is no guarantee every account can still select them. So whether 4o is usable is not decided by how well known it once was, but by how far the official retirement has progressed together with the state of your account.
The third misconception is the rule that "the bigger the version number, the higher the tier". In practice OpenAI allocates models by plan: the same version may be available on one tier and not on another. There is another easy point of confusion — names containing Fast or Ultrafast are speed tiers for allowance consumption, not new model names. OpenAI states in its Pro article that for the same model, Fast mode consumes your allowance at 2.5 times the standard rate and Ultrafast at 8 times; this is only the rate at which allowance is consumed, it does not mean the model runs faster, and it is not a separate model.
2. What differs between the models available on each plan
According to the three official articles checked on 2026-11-19, model availability by tier is as follows. This covers only whether a model is available; it does not cover prices or allowance figures for each tier, which remain as officially specified at the time.
| Plan tier | Older models (e.g. 4o) | Thinking/reasoning models | Official statement on availability |
|---|---|---|---|
| Free | Not listed individually by OpenAI | Not listed individually by OpenAI | Unlimited everyday text chat (same as Go) |
| Go | Does not include older models | Think uses GPT-5.6 Luna, does not include GPT-5.6 Sol | Unlimited everyday text chat |
| Plus | OpenAI does not state whether older models are included | The reasoning options offered within the current account are available | Wider range of models and tools than Free; refer to the model picker |
| Pro | Not listed individually by OpenAI | Includes Pro-exclusive models | Only Pro 500 includes Ultrafast (GPT-6 Astra, GPT-6.1 Sol) |
GPT-5.6 Luna and GPT-5.6 Sol in the table come from the Go article; GPT-6 Astra and GPT-6.1 Sol come from the Pro article. All are specific model names appearing in the official documentation. Cells marked "not listed individually / not stated" mean that the corresponding official document does not give an itemised list; they must not be read backwards as "definitely unavailable" or "definitely available". That is also why the model picker is the reference.
Sources by tier are as follows. The ChatGPT Go article answers directly whether Go can use older models such as 4o — it cannot; access to older models is not included in a Go subscription. Think on Go uses GPT-5.6 Luna and does not include GPT-5.6 Sol; Free and Go users alike have unlimited everyday text chat.
The ChatGPT Plus article states that Plus has broader access to models and tools than Free and can use the reasoning options offered within the current account, but it does not itemise which model names Plus includes. So judging what Plus can use still comes back to the model picker.
The ChatGPT Pro tiers article states that a Pro subscription includes Pro-exclusive models, Codex (a coding assistant) and more; among these, only the Pro 500 tier includes GPT-6 Astra Ultrafast and GPT-6.1 Sol Ultrafast — Ultrafast is offered on Pro 500 only.
In one line: older models such as 4o are being retired; Go explicitly does not include older models; Think on Go is GPT-5.6 Luna; Ultrafast is exclusive to Pro 500. Beyond that, OpenAI does not publish a fixed list of specific model names, and the model picker in your account is the reference in every case.
3. How to check which models your account can use right now
Do not rely on memorised version numbers. Check inside your own account using the steps below; the result is more accurate than any guide.
- First confirm which tier you are on: sign in to ChatGPT, click the avatar in the lower-left corner, open the account or subscription page under Settings, and look at the current plan name.
- Go back to the chat page, find the model picker (the model drop-down above or beside the input box) and see which model names are listed. Only those listed are selectable by you at this moment; anything not listed cannot be used by your account, no matter how strongly others recommend it.
- Compare against the table in section 2: see which models your tier should theoretically have, and where that differs from what the picker actually lists.
- If a model shows as "temporarily unavailable", check whether the interface gives a restoration time. OpenAI states in its Pro article that model allowances differ by tier, and that after you reach the limit for a model it may become unavailable until the allowance resets; ChatGPT will show the reset time when it is able to.
OpenAI also states that running out of allowance does not mean the subscription has ended, nor that the account has been restricted; you can switch to another available model, or wait for the allowance to reset. So when a model is greyed out, check the restoration time first, then decide whether to switch models or wait.
4. What to do when a model shows "temporarily unavailable"
Based on the answer to "Why is a model temporarily unavailable?" in the official Pro article, the order of checks is as follows.
- First confirm whether the allowance has run out: check whether a note next to the model shows an allowance limit and a reset time.
- If you need it urgently: switch to a model that is currently available and carry on; the task does not have to stop.
- If it is not urgent: wait until the reset time shown in the interface has passed, then come back to this model.
- If the model is still unavailable after the allowance resets, OpenAI advises contacting OpenAI support to verify; this is not a step a third-party top-up service can handle on your behalf.
The order in one line: check whether the allowance has hit its cap → switch if you can → if you cannot, wait for the reset → if it is still unavailable after the reset, contact official support.
5. For readers in Hong Kong and Taiwan without a foreign-currency card: how to subscribe and gain access to these models
Everything above concerns which models you can use once the account is already subscribed. For readers in Hong Kong and Taiwan, the step that usually blocks them comes earlier — payment. The ChatGPT top-up page on this site explains that official subscriptions are billed through Stripe and require a usable foreign-currency credit card; debit cards issued in mainland China, and most single-currency cards, are often declined at the payment page, and although local Hong Kong cards can often pay, the charge counts as a foreign-currency transaction and some banks apply extra risk checks to subscription merchants. Without a usable foreign-currency card the plan cannot be opened, and no additional models will appear in the model picker.
TopupCat on this site addresses exactly this payment step: you pay in RMB by scanning a QR code with Alipay or WeChat, and the ChatGPT membership is topped up to your own account, with no overseas credit card needed and no account password requested. Two boundaries should be stated up front to avoid misunderstanding:
- This site is a third-party top-up service. The Terms of Service state that membership features, allowances and available models are all subject to OpenAI's official rules at the time, and that OpenAI may adjust them at any time. The top-up solves the payment problem of not being able to get the plan opened; it does not change how OpenAI allocates models and allowances.
- This site does not promise any allowance multiplier and does not guarantee that a specific model will be available; which models you can use after activation should still be checked in the model picker using the methods in the first three sections of this article.
So the correct order is: first work out from this article which models the tier you want should theoretically be able to use, then decide whether to use a top-up. If you want to pay once by Alipay or WeChat and have the membership applied to your own account, you can start from the ChatGPT top-up plans page; verify your account before ordering and check the order page for progress after payment — the full process is in the five-step top-up guide.
Topic: Account and usage(All articles in this topic)