I understood 20× as what my $200 bought. I didn’t realize it included a temporary promotion.
The $100 / 5× and $200 / 20× comparison looked like a great deal: twice the price for four times the allowance. I can accept the $200 plan becoming less generous. Discovering that I’d remembered the bonus as the product is the annoying part.
OpenAI now says eligible Pro 200 subscribers keep their previous included allowance through October 29, 2026, while their subscription is active, then move to a lower allowance at the same $200 monthly price. Eligibility covers accounts active at any point from September 22 through September 29 at 10 a.m. Pacific Time.
Repeated promotions can teach customers to remember the bonus as the subscription. A company can keep its stated terms and still disappoint the expectations its product helped create.
The bonus became the number
The promotion was disclosed. OpenAI’s February 2 release notes introduced a limited-time doubling of Codex rate limits for Plus and Pro. The April 9 entry described the $100 tier’s standard 5× allowance, a temporary up-to-10× offer, and continuation of the $200 tier’s Codex promotion through May 31.
For the $200 tier, the historical comparison was a 10× Codex baseline doubled to 20×. That’s the promotion arithmetic, not what the original Pro launch promised in 2024. My remembered 5×/20× comparison also left out the $100 tier’s own promotion.
Baseline against baseline
Twice the price twice the allowance.
The mixed comparison I had in mind
Twice the price four times the displayed allowance.
That comparison mixes a baseline with a promotion.
The $100 tier also had a temporary up-to-10× Codex offer. This mixed 5×/20× comparison is not the complete April offer table. These are relative Codex allowances, not model-token counts, all-ChatGPT limits, or API cost.
By August 30, Thibault Sottiaux was describing the weekly 20× allowance in reassuringly literal terms:
Read original post on Xexactly what it says on the tin
A customer doesn’t rehearse the release notes every time they open Codex. They learn whether they can finish the work. Keep giving them the larger allowance, and it can become the number they judge the next bill against.
I don’t know how many subscribers made that mistake. I made it. “Temporary” was in the documentation; it wasn’t in my mental model of the subscription.
A reset starts to feel like part of the deal
The resets add another layer. These are three recent events, with September 26’s announcement and confirmation kept together. Card dates use UTC.
Banked resets for Plus, Pro, and Business accompanied the Sol/Luna launch and API price cut.
Read original post on XA reset was announced for paid Codex and ChatGPT Work accounts after a brief disruption.
Read original post on XThe announced reset had propagated. This confirms the same September 26 event, not another reset.
Read original post on XA global reset for paid ChatGPT accounts was announced after GPT-6.1 Sol’s first two days of load. The post specifies tomorrow at 10am PST.
Read original post on XThe last card is an announcement, not confirmation that the reset landed. As of this post’s October 2, 2026 research cutoff at 18:58 UTC, I hadn’t verified delivery. Its “tomorrow at 10am PST” wording is preserved from the original post; I haven’t resolved the timezone ambiguity.
Some resets followed disruption or a rough model rollout. Restoring access after trouble is useful. Still, repeated extra access can make the ordinary limit harder to remember. The next reset starts to feel like part of the deal, even when each one has a separate reason.
A refill now can change next week’s plan
A banked reset is saved for you to redeem. A global reset applies directly. The latest announcement concerns a global reset; the calendar below shows the documented rule for a full banked reset.
In OpenAI’s example, redeeming a full banked reset and resuming on Tuesday refreshes the allowance and moves the next weekly refill to around the following Tuesday. The original Friday refill no longer happens.
That doesn’t establish less usage overall. It can create a planning surprise: expect an extra Tuesday top-up and the usual Friday refill, then discover Friday’s access won’t arrive. The refill helped. The calendar assumption didn’t.
Access restored now. A new date to plan around.
This is a scheduling example, not a measured usage loss. Check Settings → Usage for the exact updated time. Source: OpenAI's banked-reset guide.
An early refill with a restarted clock is a reasonable tradeoff when you know what you’re choosing. The example doesn’t show whether the current confirmation explains it clearly. It shows why the old and new refill dates belong in that decision.
The replies don’t agree, for a reason
A partial sample of 23 public replies across three branches under the latest announcement shows how differently a reset can land. Parts of the wider thread required login; this isn’t a representative survey.
One reply and another question the benefit when an ordinary refill is near. Another points out that refill dates differ. One user says the timing works for both accounts. Those reactions can all make sense.
More control would help with planning. Saved resets with expiration dates add another calendar to manage. That reply doesn’t establish that the global reset deletes saved resets.
A reset can be generous and still confusing. Its value depends on what you have left, when the next refill arrives, and whether a saved reset is about to expire.
Three product lessons
Keep the baseline beside the bonus. Show what the monthly price buys, what the promotion adds, and when it ends. “20×” is easy to remember. Its conditions need to travel with it.
Give the refill a reason. A launch promotion, compensation for disruption, and a saved reset set different expectations. Explain the reason and the mechanism. Otherwise, customers may start budgeting around the next rescue refill.
Design the ending, too. Show the return to the baseline before a promotion expires. Show the next refill date before a banked reset is redeemed. Terms published months ago are a poor substitute for information at the moment someone makes a plan.
I’m resigned to the $200 plan being less generous. I still want to know what I can plan around. The promotion may expire on schedule; my memory of what $200 bought doesn’t come with the same timer.
Was this useful?
Thanks for the feedback.