OpenAI Doubles ChatGPT Pro Usage Limits After Backlash Over a Planned 50% Cut

Friday news brief, 30 October 2026. Summarised from reporting by BleepingComputer (Mayank Parmar, 29 October 2026), with plan-level detail from gHacks and XenoSpectrum. OpenAI has not published exact numerical limits for any tier; figures below are the multipliers and dates the company and these outlets reported.

What happened

On 29 September 2026, OpenAI emailed ChatGPT Pro 200 subscribers to say their usage allowance would be reduced after 29 October, with the planned cut widely reported as roughly half the existing allowance. The $200-a-month price was to stay the same. For a month, customers paying the highest consumer-facing price for ChatGPT expected to receive materially less for it from 30 October.

On 29 October, the day before the reduction was due to take effect, OpenAI reversed course. Instead of cutting the Pro 200 allowance, the company doubled usage limits across the Pro 100, Pro 200 and Pro 500 tiers and across Business Premium, effective immediately. Thibault Sottiaux, known as Tibo, who speaks for ChatGPT and Codex on X, confirmed the change: “We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.” He added: “We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same.”

The reversal is a direct response to customer pushback. Between the September email and the October announcement, developers and power users who rely on Codex for coding work and on ChatGPT Work for document-heavy tasks were vocal about the planned cut. The outcome is unusual: a vendor that announced a reduction and then, under pressure, moved in the opposite direction with an increase for every paid Pro tier, not just the one that complained.

What changed, tier by tier

OpenAI did not publish exact numbers, so the most useful way to read the announcement is by multiplier and by which pool of usage is affected. Pro plans have two separate pools: a weekly message limit for Chat, and a combined allowance shared between ChatGPT Work and Codex. The doubling applies to the second pool.

• Pro 100 ($100 a month). The shared Work and Codex allowance is now twice what it was. The weekly Chat limit is unchanged.
• Pro 200 ($200 a month). The planned reduction is withdrawn. Reporting indicates Pro 200 subscriptions are now unified at the previous higher allowance for both Work/Codex and weekly Chat, so customers who already had the higher level keep it and anyone who had been moved to a lower level is brought back up.
• Pro 500 ($500 a month). The shared Work and Codex allowance doubles. The weekly Chat limit is unchanged.
• Business Premium. The allowance also doubles.
• Plus. No change in this update.
• Dots. OpenAI’s always-on assistants are tracked separately. The company said the extended Dots allowance “will continue for the time being” and that Dots usage “will be managed separately from the regular allowance.”

Two details matter for anyone trying to turn this into a spending estimate. First, the doubling is of a figure OpenAI has never disclosed, so a team cannot calculate how many Codex sessions or Work documents it now gets per month; it can only observe its own consumption against the limit indicator in the product. Second, speed settings multiply consumption. According to XenoSpectrum’s plan breakdown, the Fast setting consumes about 2.5 times the allowance of Standard and Ultrafast about 8 times. A doubled pool is used up four times faster at Ultrafast than at Standard, so the headline multiplier and the practical multiplier can be very different depending on how a team has configured its tools.

Why a vendor reverses a cut

OpenAI’s reasons were not spelled out beyond “feedback”, so the following is analysis rather than reporting. Three pressures tend to push a usage-based vendor to this kind of reversal.

Churn risk at the top tier. Pro 200 and Pro 500 customers are the people who have already demonstrated willingness to pay. A cut in what they receive for the same price is the single most reliable way to make them trial a competitor, and in 2026 the competitors for coding assistants and document agents are credible. Losing even a small fraction of that cohort costs more than the compute saved by the cut.

Codex is a retention product. Developers who build Codex into their daily workflow are far stickier than casual chat users, but only while the tool is reliably available. A usage ceiling that bites mid-sprint teaches a team to keep a second tool warm. Protecting Codex usage is protecting the part of the subscription that is hardest to replace once a team has committed to it.

Capacity assumptions shifted. A reduction announced on 29 September and a doubling announced on 29 October suggest that the capacity picture, the cost per request, or both changed within a month. For customers, the lesson is not about OpenAI specifically. It is that usage limits on AI services are a dial the vendor can turn in either direction on roughly 30 days’ notice, and plans that assume a fixed allowance are plans built on a moving number.

What this means for Orange County businesses

Most organisations reading this are not individual developers deciding whether to keep a $200 subscription. They are firms where several people, sometimes several departments, have adopted ChatGPT Pro, Codex or Business Premium with varying degrees of oversight. For them, the announcement is good news on cost and a useful prompt on governance.

On cost, teams that were preparing to upgrade from Pro 100 to Pro 200, or from Pro 200 to Pro 500, to get more Work and Codex capacity may no longer need to. The doubled allowance at the lower tier may now cover the workload that was pushing them upward. That is worth checking before any renewal, because the difference between tiers across a dozen seats is real money over a year.

On governance, the episode exposes a pattern we see across Irvine, Santa Ana, Newport Beach and the wider county: AI subscriptions bought on personal or departmental cards, consumption that nobody tracks centrally, and no record of which plan terms were in force when a piece of work was produced. When the vendor changes the terms, nobody in the business knows how exposed they are until a tool stops responding in the middle of a client deliverable.

Questions to ask about your own AI subscriptions this week

1. Who holds the subscriptions, and on what payment method? If the answer involves personal cards reimbursed through expenses, the business does not control the account, cannot see usage, and may lose access when an employee leaves.
2. Which tier is each seat on, and why? After this week’s change, the reason a seat was moved to Pro 200 or Pro 500 may no longer hold. Record the reason, not just the tier.
3. What speed setting are people using? Fast and Ultrafast consume allowance at 2.5 and 8 times the Standard rate according to the reported plan detail. A team that defaults to Ultrafast is spending its doubled pool four times faster than one on Standard and may not know it.
4. Is Codex on the critical path for any client work? If a coding assistant hitting its limit would delay a deliverable, the business needs either a higher tier with headroom, a second tool, or a scheduling rule that keeps high-consumption work away from deadlines.
5. What goes into these tools? Usage limits are a cost question. What employees paste into ChatGPT Work or hand to Codex is a data-protection question. Client source code, contracts, personal data and financial records in a consumer-grade subscription are a compliance exposure regardless of how generous the allowance is.
6. Who would notice a terms change? The September email went to Pro 200 subscribers. In a business where the subscriber is a developer and the decision-maker is a director, that email may never have reached the person who owns the spend.

Consumer subscriptions versus a managed arrangement

There is a point at which a business has enough AI seats that the consumer and prosumer tiers stop being the right tool. The signals are familiar from every other category of software: more than a handful of seats, client data in the prompts, a need for audit trails, and a dependence on the tool that would hurt if it were withdrawn. At that point the conversation shifts from “which Pro tier” to whether an enterprise agreement, an API-based integration with usage you meter yourself, or a different vendor arrangement gives the business the controls it needs.

None of that is a reason to avoid the Pro tiers. For small teams and individual specialists they remain the fastest way to put capable AI tooling to work, and this week’s doubling makes them better value than they were a month ago. It is a reason to know where the line is before the business crosses it by accident.

A short review you can run before the next renewal

1. List every AI subscription the business pays for, directly or through expenses, with tier, seat holder, renewal date and payment method.
2. Pull a month of consumption from the product’s usage view for each seat and note how close each one runs to its limit at the speed setting actually in use.
3. Re-evaluate any seat that was upgraded for capacity, since the doubled allowance may have made the upgrade unnecessary.
4. Write down the tasks each seat is used for and flag any that involve client data, source code under NDA, personal information or regulated records.
5. Decide, for each flagged task, whether a consumer-grade plan is acceptable, and if not, which arrangement replaces it.
6. Set a named owner for AI subscriptions so the next vendor email about limits or pricing lands with someone who can act on it.

Firms that would like help with that review, or with moving AI tooling from scattered personal subscriptions to something the business controls, can talk to our team. Our AI services and cybersecurity services pages describe how we approach AI tool adoption for Orange County organisations.

Frequently asked questions

Did OpenAI raise the price of ChatGPT Pro?

No. The reported prices remain $100, $200 and $500 a month for Pro 100, Pro 200 and Pro 500. The change is to usage allowances, not price.

What exactly was the planned cut?

OpenAI’s 29 September email told Pro 200 subscribers their allowance would be reduced after 29 October. The reduction was widely reported as about 50%. OpenAI has not published the exact figure, and the cut has now been withdrawn.

Does the doubling apply to ChatGPT Plus?

No. Reporting indicates Plus is unchanged in this update.

Does the doubling apply to Chat messages?

Not on Pro 100 or Pro 500, where the weekly Chat limit is reported as unchanged. The doubling is to the shared ChatGPT Work and Codex allowance. Pro 200 is reported as unified at its previous higher level for both pools.

What are the new numerical limits?

OpenAI has not published them. The company described the change as a doubling of existing limits and said Pro subscribers gain capacity for Work and Codex without moving to a higher tier.

How are Dots affected?

Dots usage is tracked separately. OpenAI said the extended Dots allowance continues for the time being and is managed apart from the regular allowance.

Could the limits change again?

Yes. The September announcement and October reversal happened a month apart. Businesses should treat AI usage allowances as adjustable on short notice and plan with some headroom.

What to watch next

• Whether OpenAI publishes concrete allowance figures per tier, which would let businesses plan rather than observe.
• Whether competing coding-assistant and document-agent vendors respond with their own allowance changes.
• Any follow-up on Dots once the “for the time being” period ends.
• Whether Business Premium gains the administrative usage reporting that would make the doubled allowance measurable at the organisation level.

Technijian will update this brief if OpenAI publishes exact limits or changes the terms again.

Source: BleepingComputer, “OpenAI doubles ChatGPT Pro usage limits after backlash over planned 50% cut”, Mayank Parmar, 29 October 2026. Plan-level detail: gHacks Tech News and XenoSpectrum, 30 October 2026.