The plan that leaked as ChatGPT Pro Max is real, and it shipped as Pro 500. At DevDay on September 29, 2026, OpenAI introduced “a new $500/month plan with the highest included usage of the Pro plans and access to Astra Ultrafast in ChatGPT Work and Codex,” in its release notes. Its DevDay recap puts the allowance at “25 times the ChatGPT Plus allowance.”
This page opened on September 26 as a tracker of a leak. It is now the record of what shipped, graded against what the leak said, read from OpenAI’s release notes, its Pro tiers help article, the DevDay recap and the pricing page.
On this page · 9 sectionsOpen
- OpenAI launched Pro 500 on September 29, 2026, at DevDay: $500 a month, the highest included usage of the three Pro plans, and access to Astra Ultrafast in ChatGPT Work and Codex.
- It is the plan that leaked as Pro Max. The $500 price, the faster Work and Codex and the higher usage all shipped; the name did not.
- OpenAI’s DevDay recap puts the Pro 500 allowance at 25 times the ChatGPT Plus allowance.
- Ultrafast offers up to 8x faster generation in Codex (300 tokens per second) and up to 6x in the API. Among Pro plans it is available only on Pro 500.
- Pro 200 is open to new subscribers again, with a lower allowance; existing subscribers keep their previous allowance through October 29, 2026.
- Pro 500 is sold through ChatGPT on the web at launch.
§ 01The leak, graded
| Claim | Outcome |
|---|---|
| Priced at $500 a month | Confirmed: Pro 500 is $500 a month |
| Fastest Work and Codex | Confirmed: Astra Ultrafast in ChatGPT Work and Codex |
| Higher usage limits | Confirmed: the highest Pro allowance, 25 times Plus |
| Launches at DevDay | Confirmed: announced September 29 |
| Named Pro Max | No: OpenAI calls it Pro 500 |
| A promax plan type in OpenAI’s Codex code | The code was real; the public name is Pro 500 |
| Runs on Cerebras capacity | Still unconfirmed; OpenAI’s pages do not say |
Four of the leak’s claims held and one did not. The name is the miss: every OpenAI page we read uses Pro 100, Pro 200 and Pro 500.
§ 02What Pro 500 includes
- The highest included usage of the three Pro plans; OpenAI’s recap puts it at 25 times the ChatGPT Plus allowance.
- Astra Ultrafast in ChatGPT Work and Codex. “Among Pro plans, Ultrafast is available only on Pro 500.” It uses the plan’s included usage first, then the credit balance.
- The Agents API with computer use in Codex and ChatGPT Work, which the recap lists for Pro 500 and Enterprise.
- Everything in Pro, including dot, the always-on agent that the pricing page now lists on the Pro card.
“Pro 500 subscriptions are available through ChatGPT on the web at launch.”
§ 03The Pro ladder now
| Plan | Price | Ultrafast |
|---|---|---|
| Free | $0 | No |
| Go | $8 | No |
| Plus | $20 | No |
| Pro 100 | $100 | No |
| Pro 200 | $200 | No |
| Pro 500 | $500 | Yes |
The pricing page still shows one Pro card, “From $100,” with a choice of three usage tiers. Pro 500 is the first ChatGPT plan above $200.
§ 04Pro 200 reopened, with a smaller allowance
On September 10, OpenAI paused new sign-ups to the $200 plan. With Pro 500, it reopened them: new Pro 200 subscriptions are available again at $200, but subscriptions that are not eligible for grandfathering get a lower included allowance. Subscribers whose Pro 200 plan was active at the eligibility cutoff, or in the seven days before it, keep their previous allowance through October 29, 2026, then move to the lower one at the same price.
§ 05Ultrafast, in numbers
OpenAI calls Ultrafast its premium speed tier. The recap gives up to 8x faster token generation in Codex, 300 tokens per second, and up to 6x in the API. GPT-6 Astra Ultrafast is available now in the API and in ChatGPT Work and Codex on Pro 500 and Enterprise; GPT-6.1 Sol Ultrafast is coming soon. “At launch, buying credits on Pro 100 or Pro 200 does not unlock Ultrafast.”
§ 06What we are watching for
- GPT-6.1 Sol Ultrafast, which OpenAI lists as coming soon.
- Pro 500 purchases outside the web, which OpenAI’s notes limit to ChatGPT on the web at launch.
- October 29, when grandfathered Pro 200 subscribers move to the lower allowance.
- Any statement on the hardware behind Ultrafast; the Cerebras link remains press inference.
§ 07The record
As of September 30, 2026: Pro 500 is live at $500 a month with the highest Pro allowance and Astra Ultrafast. It is the plan that leaked as Pro Max, under a different name. Our full DevDay record is at OpenAI DevDay 2026.
§ 08Update log
- September 30, 2026: flipped from tracker to release record. OpenAI launched the plan as Pro 500 at DevDay on September 29; leak graded above.
- September 26, 2026, 04:50 UTC: page opened as a tracker of the leaked Pro Max plan.
§ 09Sources
- OpenAI, ChatGPT release notes, September 29, 2026 entry, read September 30.
- OpenAI, About ChatGPT Pro tiers, read September 30, 2026.
- OpenAI, DevDay 2026 recap, September 29, 2026.
- OpenAI, ChatGPT pricing, read September 30, 2026.
- openai/codex, pull request 47971, merged September 25, 2026, 00:20 UTC.
- TestingCatalog, OpenAI prepares new $500 per month Pro Max plan, September 24, 2026.
Q1How much is ChatGPT Pro 500?
$500 a month, per OpenAI’s release notes and its Pro tiers help article. The other Pro plans are $100 and $200.
Q2Is Pro 500 the same as the leaked Pro Max?
It matches the leak on price, speed and usage. OpenAI’s pages call it Pro 500; none of them use the name Pro Max.
Q3What is Ultrafast?
OpenAI’s premium speed tier: up to 8x faster token generation in Codex, about 300 tokens per second, and up to 6x in the API. It uses your included usage first, then your credit balance.
Q4Can I still buy Pro 200?
Yes. New Pro 200 subscriptions reopened with a lower included allowance at the same $200 price. Subscribers active at the eligibility cutoff keep their previous allowance through October 29, 2026.
Q5Does buying credits on Pro 100 or Pro 200 unlock Ultrafast?
No. OpenAI says credits on those plans do not unlock Ultrafast at launch; among Pro plans it comes only with Pro 500.
Q6Is $500 a month worth it for agent work?
It depends on whether the bottleneck is speed or ownership. Pro 500 makes one person’s Codex and Work runs faster. An AI employee platform such as CellCog prices the work instead of the seat: each employee has its own inbox, task board and memory and carries a role from one session to the next.