ChatGPT Pro comes in three tiers: 100, 200, and 500. If you mainly use Codex, the 100 is the choice if one subscription is enough, consolidating into the 200 is the candidate if you hold two 100 subscriptions, and the 500 is the candidate if even the 200 isn't enough or you need Astra Ultrafast. Since the October 29, 2026 increase, the Work and Codex allowance, with Plus as 1, is 10 for the 100, 20 for the 200, and 50 for the 500 (calculated by this site from OpenAI's announcements). Two 100 subscriptions and one 200 have the same total allowance, and at the official US dollar prices both cost $200 per month. Local-currency prices vary by country or region, so compare the amounts on your own pricing page.
Work and Codex allowance guide, and reasons to choose
The baseline for people who need one subscription. You don't need to move to a higher tier to use Sol.
The consolidation option for people with two 100 subscriptions. You can put the standard allowance into one account.
For ongoing allowance shortfalls or Astra Ultrafast. Looking only at standard allowance, the US dollar unit price is the same as the 200.
The multiplier baseline is Plus (unchanged by the October 29 change). These are not multipliers for message counts, completed tasks, or image counts, and they are also separate from the per-model limits in regular Chat.
What you'll learn in this article
- How much is known about the allowance multipliers
- A detailed comparison of prices and features
- Is it better to consolidate two 100s into one 200?
- How Fast and Ultrafast change how much you can use
- Reset passes, paid resets, and credits: the differences
- Choosing based on how you use it
- What to check before changing your plan
- Conclusion
- FAQ
- Related articles
How much is known about the allowance multipliers
With the September 29, 2026 announcement, Pro 500 was added, and Pro 200, which had stopped accepting new sign-ups, reopened. On October 29, the Work and Codex allowances of the 100 and the 500 were each doubled, and every 200 subscription was brought to the same (higher) allowance. The official Pro help page explains the monthly prices of the three tiers, their relative allowance sizes, and this change, but it does not give multipliers relative to Plus. The multipliers as of September 29 came from a staff announcement for the 100 and 200, and from OpenAI's own announcement for the 500.
Plus = 1 → Pro 100 = 10 → Pro 200 = 20 → Pro 500 = 50
The 200 is twice the 100, and the 500 is five times the 100 and 2.5 times the 200. The ratios among the three are the same as on September 29.
What's being compared here is the relative size of usage allowances for Work, Codex, and similar features. The official pricing document states that Work and Codex share the same allowance, and that consumption varies with the model, reasoning effort, input and output, tool use, caching, and more. "Since 100 handled 100 tasks, 200 will handle 200" does not hold. Even on the same plan, longer conversations or heavier research will change how quickly your allowance runs down.
Pro 200's allowance became a single allowance on October 29
As of September 29, a newly started 200 subscription was 10x Plus, while existing 200 subscriptions were to keep their higher allowance until October 29 and then move to the new allowance. On October 29, OpenAI dropped this cut: it doubled the 200 subscriptions that were on the lower allowance, so every 200 now has the same allowance. Those already on the higher allowance stay where they were. The price is still $200 per month, cancelling and resubscribing gets you the same allowance, and Ultrafast is not included (official help, release notes). The Plus allowance has not changed.
Search results still say "the 200 is 10x Plus," but that was the allowance of 200 subscriptions started before October 29. Today every 200 has the same allowance, which is 20x Plus by the calculation above. You no longer need to compare differently depending on when you subscribed.
From December, the ChatGPT and Codex usage allowance on Plus, Pro, and Business is scheduled to become a single shared allowance (official help, checked November 3, 2026). Usage will be measured in tokens, and how fast it goes down depends on the model you choose and the task. Instant chat does not draw from this allowance. The amount each plan gets after the change has not been announced yet, so read the multiples in this article as figures for the Work/Codex allowance before the December change.
A detailed comparison of prices and features
The official US dollar monthly prices are $100, $200, and $500. Prices in local currency, including tax, vary by country or region, so check the amount shown on your pricing page.
| Comparison item | Pro 100 | Pro 200 | Pro 500 |
|---|---|---|---|
| Official monthly price | $100 | $200 | $500 |
| Price in local currency (incl. tax) | Varies by country or region | Varies by country or region | Varies by country or region |
| Monthly price × 12 | $1,200 | $2,400 | $6,000 |
| Work/Codex allowance / multiple of Plus (calculated) | 10x | 20x | 50x |
| GPT-6.1 Sol | Available | Available | Available |
| Standard / Fast | Available on eligible models | Available on eligible models | Available on eligible models |
| Astra Ultrafast | Not included | Not included | Included |
| Work/Codex 5-hour limit | None in current Pro guidance | Same | Same |
| Work/Codex weekly limit | Possible / check your usage | Same | Same |
| Pro model allowance in regular Chat | GPT-6 Pro and 5.6 Sol Pro share a weekly allowance | GPT-6 Pro weekly allowance, 5.6 Sol Pro daily allowance, and others | Per-model figures not disclosed |
| Storage | 100 GB | 100 GB | 100 GB |
Sources: Pro plans, Work/Codex pricing, speed specifications, Pro model limits in Chat. The 12-month figures are calculated values and do not represent annual plans.
100 GB is file storage capacity, not a monthly token count or Codex workload. Pro also includes image generation and deep research, but the 10/20/50 ratios can't be applied to the number of images or regular Chat questions. If you mainly use Chat, check the limits and reset times on the model selection screen before you choose.
In regular Chat, the two models available on the 100 share a weekly allowance, while the 200 has both a daily and a weekly allowance. When you use up GPT-6 Pro, you switch to GPT-5.6 Thinking Medium. You can manually select 5.6 Sol Pro as long as that model's daily allowance and the combined daily allowance for both models remain. The help page published after October 7, when GPT-6 became the default in chat (verified October 12, 2026), still lists GPT-5.6 Thinking Medium as the fallback.
Is it better to consolidate two 100s into one 200?
At the official US dollar prices, two 100s cost $200 per month, the same as the 200. The Work and Codex allowance is also the same: 10 + 10 = 20 and 20. The difference is not the price but whether the allowance sits in one account. Local-currency prices vary by country or region, and in some regions the 200 costs less than two 100s, so compare the billed amounts on your own screen. For Codex-centered use with the same model, reasoning effort, and speed, consolidating into the 200 is the first candidate.
Same allowance, same US dollar price. The difference is the number of accounts.
Balances, history, and billing are separate.
The same amount of allowance, usable in one account.
In US dollars, there is no price difference. Local-currency prices vary by country or region, and the comparison excludes temporary benefits.
However, the allowances of two subscriptions can't be combined into a single chat. As the official explanation of account switching states, history, memory, and billing are independent for each account, and consolidating into the 200 won't merge the history from the other account. If you keep both subscriptions, since October 30, 2026 you can also stay signed in to two accounts and switch between them in the mobile app (release notes). If you want to separate work and personal use, weigh that convenience together with the cost difference.
OpenAI also limits how many paid subscriptions can be tied to the same payment card or device (card and device limits, checked November 3, 2026). If too many subscriptions use the same card, new purchases are blocked, and there is also a cap on how many paid accounts you can use on one device. Past that cap you see "Device limit reached" and can't use that account on that device. The limit covers Go, Plus, and Pro 100, 200, and 500, and the help page doesn't give the actual numbers. Subscriptions bought before the limits took effect may continue, but after cancelling there is no guarantee you can subscribe again the same way. Keep this in mind when deciding whether to keep two subscriptions or consolidate into the 200.
Looking only at standard capacity, 500 is not cheaper than 200
In US dollars, 100/10, 200/20, and 500/50 each come to $10 per allowance unit, so the unit price is the same. Moving up to the 500 does not lower the unit price of standard capacity. In local currency, the price ratio between tiers can differ by country or region, so divide the amounts on your pricing page by 10, 20, and 50 to check.
That said, if having a large allowance on one account, or shortening wait times with Astra Ultrafast, is valuable to you, the 500 is a candidate. Conversely, if Sol at standard speed is enough for you, you don't need to choose the top tier just because it's the top tier. For how to choose models and reasoning settings, see GPT-6.1 Sol vs. GPT-6 Sol.
How Fast and Ultrafast change how much you can use
Even on the same plan, changing the speed setting changes how much allowance you consume. The official pricing document shows different multipliers for in-plan allowance and additional credits. Fast is not limited to the 200 and above; it is also available on the 100 for eligible models.
| Relative to Standard on the same model as 1 | In-plan allowance usage | Credit usage | Decision point |
|---|---|---|---|
| Standard | 1x | 1x | Baseline when comparing standard capacity |
| Fast | 2.5x | 2x | Available on eligible models / use for urgent work |
| Astra / GPT-6.1 Sol Ultrafast | 8x | 6x | Pro 500 only for individuals / check availability |
With the same Astra and the same token mix, if all of the 500's usage is in Ultrafast, dividing the 50 allowance by the 8x consumption multiplier gives 6.25. That is equivalent to 62.5% of the 10 from Standard on the 100.
Astra Ultrafast's "up to 8x" refers to the speed of token generation compared with Standard on the same model. It does not mean the entire task, including research, building, and browser operation, finishes 8 times faster. The official Pro help page and the speed specifications (verified November 2, 2026) list Astra and GPT-6.1 Sol as Ultrafast-eligible. For individuals, only Pro 500 is eligible. The 100 and 200 can't use it even if you buy additional credits. Enterprise and Edu depend on the contract and admin settings.
To test speed, record both the time to completion and how fast your allowance drains on your usual work. Comparing Astra Ultrafast with Sol Standard also mixes in model differences, so compare on the same model. You can also use Standard normally and switch to Fast only for work with an approaching deadline.
Reset passes, paid resets, and credits: the differences
Even when your allowance temporarily recovers, don't count it toward your monthly contracted capacity. Scheduled renewals, free resets from OpenAI (global resets, which apply automatically to eligible accounts, and banked reset passes, which you use yourself), paid instant resets, and additional credits are separate mechanisms, and which one you use changes when your allowance next renews. "Global" in global reset does not mean every account or every limit (official help).
You use a distributed pass yourself, once, before it expires. Renews the target allowance and changes the weekly renewal date. Not always distributed again.
Applied immediately upon purchase, even if you have allowance left. Cannot be saved. A new one-week period starts with your next Work/Codex request.
A balance that covers additional use of supported features after your in-plan allowance is used up. It does not remove all the limits in regular Chat.
Official explanations: banked reset passes / paid weekly resets / credits for personal plans.
Using a banked reset pass doesn't keep your original weekly renewal date
According to the official explanation of banked passes, if your renewal is scheduled for Friday but you use a banked reset pass on Tuesday, the next renewal comes around the following Tuesday, not on the original Friday. Paid resets work the same way: the next automatic renewal is seven days after your first Work/Codex request following the purchase. Neither one adds a separate allowance on top of your current balance.
What if the automatic Friday renewal is one day away?
- Thursday: the target allowance recovers
- Friday: the original renewal is gone
- Next renewal: around the following Thursday
You lose the next day's automatic recovery and one pass. It's only worth using when a deadline today or an expiring pass makes it necessary.
- Thursday: wait on work that falls short of allowance
- Friday: the regular allowance recovers without using a pass
- Pass: saved for the next shortage before it expires
Your regular allowance recovers, and you keep the pass for the next shortage.
If it's not urgent and the pass won't expire, waiting for the automatic renewal is recommended.
Passes are basically for when you truly run short on necessary work and can't wait for the next automatic renewal. Using up your allowance on non-urgent work right after an automatic renewal, then immediately using a pass, is wasteful. Pro has no 5-hour limit, so what a banked reset pass (shown on screen as a "full reset") restores is the weekly allowance. On Plus, it covers both the 5-hour and the weekly allowance. On every plan, an allowance that is still at 100% does not restart, and its renewal date does not change.
How to use a banked reset pass, and the conditions
In the desktop app, open Settings → "Usage & billing" → "Usage limit resets" and click "Use reset." On the web you can use a pass from the usage settings, and in the Codex CLI from "Redeem reset" in /usage; you can't use one in the mobile app. "Buy an instant reset" on the same screen is a paid reset and does not use up a pass (official help, verified November 2, 2026).
- You can use a pass only when at least one of the limits it covers has less than 80% remaining. If all of them are at 80% or more, you see "Your usage does not need a reset right now," and the pass stays unused.
- If you turn on "Allow your assistant to use resets," you can ask Codex in a conversation to use a pass for you. This works only when your 5-hour or weekly allowance has 10% or less remaining, and passes are never used without your request. If you want passes used only by hand, turn it off.
- A banked reset pass is used up only when at least one target usage allowance is actually restored.
- If a global reset arrives after you've used a pass, your allowance does not stack above 100%, and the pass you used is not returned. Passes you haven't used keep their original expiry date.
- The "History" tab of "Usage limit resets" shows the passes you received and used in the past 30 days. Global resets and paid resets are not listed.
- Passes are not cash or API credits, and support does not extend expiry dates or reissue passes that expired or were used by mistake.
The fine print for paid resets and credits is as follows.
- Paid resets are generally non-refundable, and availability depends on the account and billing country. The purchase option does not appear while the target allowance has not decreased at all.
- The credit balance is separate from the ChatGPT wallet (a US dollar balance such as gift cards). If work uses up the balance at the same time, it goes negative, and your next purchase or auto top-up first covers the shortfall.
- The monthly cap for auto top-ups applies only to auto top-ups, not to individual purchases. Buying credits on the 100 or 200 still doesn't unlock Ultrafast.
- Credits are valid for 12 months from purchase, and any unused amount expires and does not carry over. With the same account, they can be used across Codex, ChatGPT Work, Word, Excel, and PowerPoint. The setting that lets partner tools you signed in to with ChatGPT use your plan allowance or credits is off at first and applies only after you turn it on yourself (official help, verified October 19, 2026).
- If you're signed in with a ChatGPT account, the safety checks for Work/Codex automatic review are free and don't reduce your usage allowance (official help, verified October 12, 2026).
Credit and paid reset prices are shown on your account screen.
Choosing based on how you use it
First, use one normal week with the same model, reasoning effort, and speed, and see which tasks run short. Don't decide your monthly capacity based only on a week with a temporary reset or a week right after switching to a new model. If one 100 is enough, it has the lowest fixed cost among the Pro plans. If you're not sure you need Pro at all, include Plus as a candidate. Before moving to a higher plan, you can also move well-scoped tasks to GPT-6.1 Terra. OpenAI says it gets far more work done than GPT-6.1 Sol or GPT-6 Astra within the same allowance. We compare them in "What Is GPT-6.1 Terra? Pricing, Performance, and How to Choose Between It, Sol, Astra and Claude".
| Usage pattern | First option to consider | What to check before deciding |
|---|---|---|
| One 100 has allowance left over every week | One 100 subscription / consider Plus if needed | Pro models or other features you need, and typical weekly consumption |
| Using two 100s, mainly for Codex work | Consolidate into the 200 | Actual billing for both subscriptions, benefits, change date, and whether you need to keep history separate |
| Only a few days a month run short | 100 or 200 + additional usage | Price on the purchase screen, remaining balance, and next renewal date |
| Even the 200 runs short every normal week | Compare the 500 and 200 + additional usage | Whether additional costs above $300 per month would continue |
| Want to greatly reduce waiting for Astra generation | 500 Ultrafast | 8x consumption and how much time you save across the whole task |
| Mainly using Pro models in regular Chat | Check the Chat-side limits before choosing | Per-model allowances and reset times, not the Work/Codex multipliers |
How much additional usage can you buy before it costs more than now?
At US dollar prices, two 100s cost $200. If you're not buying extra now, consolidating into the 200 keeps the fixed cost at the same $200, so any additional usage you buy is added to what you pay now. With the option of a single 100, the total stays at $200 or less as long as additional spending is $100 or less per month. If you pay in another currency, recalculate with the amounts shown on your screen. Of course, this assumes you can finish the same work when you need to.
200: $200 + that month's additional cost
Current 2 × 100: $200 + that month's additional cost
500: $500 + that month's additional cost
Don't treat free temporary benefits as monthly income. Count them separately only in months when you could use them.
The 500 and "200 + additional usage" come out to the same amount when additional costs are $300 per month. Equal amounts don't mean equal capacity, so compare the consumption rates for each model and speed setting, along with the features only the 500 offers.
What to check before changing your plan
You can change your plan from ChatGPT's "Settings → My plan." Your remaining balance is under "Settings → Usage" or the Codex usage screen (display names may vary by device and language). The official help says to cancel through the route you purchased from, so first confirm whether your subscription is a web subscription or was made through Apple or Google.
- Check the actual billed amount for each of the two subscriptions. If they're billed through different channels or at different prices, recalculate instead of assuming $200.
- Make sure you're looking at the same account's balance. If the browser and the Codex app show different figures, check that both are signed in to the same account.
- Record banked reset passes and additional credits separately. You can see pass expiry dates on the "Usage limit resets" screen. If you don't know how they're handled after a plan change, check on screen or with support.
- Read the change effective date, first charge, and prorated amount on the screen just before you purchase. Depending on how proration works, the first month's charge may differ from the regular monthly price.
- Confirm that renewal is stopped for any subscription you no longer need. Deleting the app does not cancel a subscription. Even after cancelling, history from another account won't move to the new subscription.
Official sources say support cannot reset the ChatGPT or Codex limits themselves. However, they can investigate counting errors and problems such as being unable to use a feature after renewal. Authentication errors and "thread not found" are not signs of insufficient allowance. If you want to organize your model settings, you may also find how to change the model and reasoning effort in chat helpful.
Conclusion
If you're running two 100 subscriptions mainly for Codex, consolidating into the 200 is a strong option that brings the same total allowance into one account. At US dollar prices, the fixed cost stays at $200, and where the local price of the 200 is lower than two 100s, it also lowers your fixed cost. If one 100 is enough, choose the 100. If you keep running short even on the 200, or need Astra Ultrafast, consider the 500.
What determines the decision is the amount actually billed in your currency, the model, reasoning, and speed you use, your typical weekly consumption, and the price of additional purchases. Since October 29, every 200 has the same allowance, and the three tiers stand at 10, 20, and 50. Count free resets and passes separately, as a bonus only in the months you could use them.
FAQ
Does Pro 200 give twice the usage of the 100, and the 500 five times?
On the Work and Codex allowance with Plus as the baseline (10, 20, and 50), yes, that is the relationship. It is not a multiplier for the number of questions, images, or completed tasks. Compare under the same model, reasoning effort, and speed, and check the limits for regular Chat models separately.
Do you need the 200 or 500 to use GPT-6.1 Sol?
No. You can use it on the 100, and 6.1 Sol is also available on Plus. The reasons to choose a higher tier are the size of the allowance and features such as Ultrafast. There is no official explanation that a higher tier gives better answers with the same model and reasoning settings.
If there are surprise resets, is a cheaper plan the better deal?
Don't base your monthly costs on this, because future distributions are not guaranteed. The 28-day campaign in October ended on November 1 (UTC), and even during it, a reset or pass distribution came on only 14 of the days. Spending for a month with a banked reset pass goes down, but record it separately from normal renewals, paid resets, and additional credits.
Does Pro come with API credits?
It does, since October 28, 2026: $100 a month on Pro 100, $200 on Pro 200 and $500 on Pro 500. You claim them under Settings → Billing once 7 days have passed since you subscribed. They expire each month without rolling over, and they can't be used for Codex running on an API key. They also don't add to your ChatGPT allowance or to the allowance for Codex used while signed in (OpenAI's help center). Conversely, Codex additional credits belong to the ChatGPT side and can't be used on the API. To compare with Claude's Max or Google AI Pro and Ultra, see "Do Subscriptions Include Free API Credits? Claude, Gemini and ChatGPT Plan Credits Compared".
Related articles
- Comparing image generation in ChatGPT, Codex, and Claude: allowance consumed by image generation and ChatGPT's image limits
- How to use dot, its pricing, and how it differs from Codex: how continuous agents relate to usage allowances
- How to fix Codex capacity errors: when "Selected model is at capacity" appears even though allowance remains
- How to fix ChatGPT responses that stop midway (Message delivery timed out): how to tell a weekly allowance limit apart from connection or request limits
- How to use Codex Security, its pricing, and how to read the results: dedicated security scanning
- How to run tasks on a schedule in ChatGPT: the number of scheduled tasks and the usage allowance for the tasks they run
- How to use ChatGPT for Word, and its pricing: editing in Word also uses the shared Plus/Pro allowance