If you pay OpenAI $200 a month for ChatGPT Pro, your plan is about to get a lot less generous. Emails sent to subscribers after last week’s DevDay say that from October 30, 2026, the included usage on the $200 Pro plan drops from 20 times to 10 times the ChatGPT Plus allowance in ChatGPT Work and Codex, while the price stays exactly the same. Alongside the cut, OpenAI has introduced a brand new $500 a month Pro 500 tier and reopened the $200 plan to new buyers at the smaller allowance.
This is one of the sharpest subscription downgrades in the AI industry’s short history, and it tells a bigger story about where frontier AI pricing is heading. Flat-rate plans are being re-priced to match the real cost of serving models, and the era of one fixed fee for as much as you can use is closing.
What exactly changes on October 30
According to the subscriber email, which was posted in full on OpenAI’s community forum and reported by MadRobot, mixed-news and ExplainX, three things happen to ChatGPT Pro 200 on October 30:
- Included usage halves: ChatGPT Work and Codex usage falls from 20x to 10x the ChatGPT Plus allowance.
- Chat messages halve: the GPT-6 Pro message limit in Chat drops from 200 to 100 messages per week.
- The price does not move: it stays at $200 a month (about Rs 17,600).
Existing subscribers keep their old 20x limits through October 29, 2026, and OpenAI is softening the change with a one-time grant of 62,500 usage credits, valued by the company at $2,500, that expire on December 31, 2026. New sign-ups after the reopen sit on the smaller formula immediately. OpenAI’s explanation, quoted from its help center, is that the lower allowance reflects “our increasingly efficient models”, because newer models do more with less.
The new Pro 500 tier and its Ultrafast perk
The email also announced Pro 500, a $500 a month plan (about Rs 44,000) that OpenAI calls its highest-usage tier. It includes 25x the Plus allowance, the first Dot included like the other Pro plans, and one exclusive feature: Ultrafast, described as “a faster service tier for GPT-6 Astra”, chosen in the model picker and available in ChatGPT Work and Codex. Buying credits on Pro 100 or Pro 200 does not unlock Ultrafast at launch, and Pro 500 is sold only through ChatGPT on the web for now.
The full lineup now reads like this:
| Plan | Price per month | Work and Codex allowance | Cost per unit of usage |
|---|---|---|---|
| ChatGPT Plus | $20 | 1x | $20.00 |
| ChatGPT Pro 100 | $100 | 5x | $20.00 |
| ChatGPT Pro 200 (from Oct 30) | $200 | 10x | $20.00 |
| ChatGPT Pro 200 (until Oct 29) | $200 | 20x | $10.00 |
| ChatGPT Pro 500 | $500 | 25x | $20.00 |
That table is the real story. Every tier now prices out at the same $20 per unit of Plus-equivalent usage. The old $200 plan was the only one that offered a true volume discount, at $10 a unit, and it is gone. Buying a bigger plan now gets you features and speed, not cheaper work.
What Pro subscribers should do before October 30
First, check whether you received the email. OpenAI says only affected subscribers get one, and the 62,500 credits will show in Settings, then Usage, if they apply to you. Second, audit your actual usage: if your heaviest weeks land well under 10x the Plus allowance, the change barely touches you, and the free credits will cover the gap past the cutover. If you are consistently near the old 20x ceiling, you have three options before October 30: push heavy workloads forward, move to Pro 500 for 25x and Ultrafast, or price the same work on the API, where GPT-6.1 Sol costs just $2 per million tokens.
One commitment in the email is worth noting for power users: OpenAI says it will not reintroduce the old 5-hour usage window, so Pro usage stays a weekly budget you can spend in one sitting. And the new Dots, the always-on agents announced at DevDay, are included on Pro plans, with work the Dot does directly not drawing the ChatGPT allowance.
The flat-rate era is ending across the industry
Zoom out and this looks less like an OpenAI whim and more like the industry finding its real price. We priced every frontier model per million tokens this month, and the pattern is the same at the API layer: vendors cutting prices as models get more efficient, then re-pricing subscriptions so the margins hold. Anthropic cut Claude Opus 5.5’s price by 20 percent at launch, and Google opened a new pricing tier for Gemini 4 Argon. The $200 plan’s 20x allowance was, in effect, a subsidy printed when models were expensive to serve per useful hour. OpenAI has decided to stop printing it.
For subscribers, the math is now simple. Re-price your seat before October 30, use the credit grant before December 31, and remember the new rule of ChatGPT pricing: bigger plans buy speed and features, not discounts.
