OpenAI used DevDay 2026 on September 29 to reshape its top ChatGPT subscription. It added a third Pro tier, Pro 500, at $500 a month. It reopened the $200 Pro plan, which had been closed to new sign-ups since September 10, but with half the usage it used to carry. Existing $200 subscribers keep their current limits until October 29 and then move to the smaller allowance on October 30, with a one-time credit grant to soften the change.
If you pay for ChatGPT Pro, or you are deciding whether to, this matters because the value per dollar on the $200 plan halves overnight while the price stays the same. This guide covers what Pro 500 includes, exactly what changes on Pro 200, what Ultrafast is and how fast it burns your allowance, a full plan comparison, how Claude Max and SuperGrok Heavy stack up, and a plain recommendation on whether to upgrade, stay or downgrade.
What is ChatGPT Pro 500?
Pro 500 is the highest individual ChatGPT plan. Per OpenAI's DevDay announcements as reported by Engadget, The Next Web and dev.to, it includes:
- Price: $500 per month before tax ($6,000 over a year).
- Usage: 25 times the ChatGPT Plus allowance, the largest of any self-serve plan. The allowance is shared between Codex and ChatGPT Work.
- Ultrafast: access to GPT-6 Astra Ultrafast in Codex and ChatGPT Work. Among the Pro plans, only Pro 500 has it. Buying extra credits on Pro 100 or Pro 200 does not unlock it.
- A dot: one always-on agent (OpenAI calls them "dots"), included on every Pro tier. Conversations with your dot do not count toward ChatGPT usage limits.
- Everything else in Pro: Pro models, Codex, deep research, image creation, memory and file uploads, according to OpenAI's Help Center article on Pro tiers.
OpenAI's Codex lead Thibault Sottiaux framed the reworked plans around API-equivalent value: Yahoo Tech quotes him saying "the changes net out at half the dollar in API spend compared to the prior plan." Pro 500 is not a volume discount. It is the same rate per unit of Plus usage as every other paid tier, plus the Ultrafast gate.
What changes on the $200 ChatGPT Pro plan?
This is the part most existing subscribers are searching for. Starting October 30, 2026:
- Codex and Work usage drops from 20x to 10x the Plus allowance.
- GPT-6 Pro chat messages drop from 200 to 100 per week.
- Price stays at $200 a month.
- The five-hour limit is not coming back. Usage stays on a weekly budget, according to The Next Web.
New subscribers who sign up now get the 10x allowance immediately. OpenAI's Help Center says new Pro 200 subscriptions "that aren't eligible for grandfathering" include the lower allowance.
What is the $2,500 credit for existing Pro subscribers?
Existing $200 subscribers keep 20x until October 29 and receive a one-time grant of 62,500 credits, worth about $2,500, that expires on December 31, 2026. The Next Web describes it as a usage credit, and the credit figures and expiry date are repeated in the OpenAI Developer Community thread on the change. These are the ChatGPT/Codex usage credits that let you keep working after your plan allowance runs out. None of the sources we could open describe them as API platform credits, so do not plan on spending them in the OpenAI API.
Two months of extra headroom is useful, but it is not a permanent fix. After January 1, 2027, a grandfathered Pro 200 subscriber is on exactly the same 10x plan as a new one.
Why are people angry about the Pro 200 cut?
Engadget's headline called it a nerf, and the reaction on Reddit and the OpenAI community forum was sharply negative. The forum thread asking OpenAI to grandfather existing subscribers includes lines such as "Same price, half the product" and "A one-off credit that expires in two months does not compensate for permanently halving the thing I'm paying for every month." MakeUseOf described the credit grant as damage control after the backlash. The timing added to it: Sottiaux posted the change ahead of the keynote, saying he wanted to "share this change ahead of time so you can all understand it before we shower you with good news."
ChatGPT plans compared (2026)
Here is every current ChatGPT plan with only the values we could confirm. "Not published" means we did not find an official figure.
| Plan | Price | Usage vs Plus | GPT-6 Pro messages | Ultrafast | Dot included |
|---|---|---|---|---|---|
| Free | $0 | Below Plus (not published as a multiple) | No | No | No |
| Go | $8/mo (US) | Not published as a multiple of Plus | No | No | No |
| Plus | $20/mo | 1x (baseline) | Not published | No | No |
| Pro 100 | $100/mo | 5x | Not published | No | Yes |
| Pro 200 (grandfathered, until Oct 29) | $200/mo | 20x | 200/week | No | Yes |
| Pro 200 (new subs, and everyone from Oct 30) | $200/mo | 10x | 100/week | No | Yes |
| Pro 500 | $500/mo | 25x | Not published | Yes (GPT-6 Astra) | Yes |
| Business Standard | $25/user/mo ($20 annual) | Not published | Not published | No | No |
| Business Premium | $125/user/mo ($100 annual) | 5x Business Standard | Not published | Not published | Yes |
| Enterprise | Custom | Custom | Not published | Yes | Beta (admin must enable) |
Two things jump out. First, every Pro tier now costs the same $20 per "Plus unit": $100 for 5x, $200 for 10x, $500 for 25x. There is no bulk discount for going bigger. Second, Pro 200 is now exactly two Pro 100s. The only reasons to pick the $500 tier over the $200 tier are the extra 15x of headroom and Ultrafast.
The Pro 100 tier launched in April 2026 with 5x the Plus Codex allowance (Thurrott). Business Premium seats were announced in August 2026 with a waitlist, and the DevDay coverage says they include a dot as well.
What is Ultrafast, and how fast is it?
Ultrafast is a premium speed tier for GPT-6 Astra. OpenAI's figures, as reported from the DevDay recap:
- Up to 8x faster token generation in Codex, reaching up to 300 tokens per second.
- Up to 6x faster in the API.
- In ChatGPT, available only on Pro 500 and Enterprise, in ChatGPT Work and Codex.
- A GPT-6.1 Sol Ultrafast variant is due "in the coming days," with no price published yet.
"Up to 8x" refers to token generation speed. Tool calls, test runs and file I/O inside an agent loop do not get faster, so a long Codex task will not finish 8x sooner end to end.
How much does Ultrafast cost?
In the API, GPT-6 Astra Ultrafast costs six times the standard rate. Our GPT-6 Astra pricing breakdown covers the standard tier in detail.
| Model / speed | Input per 1M tokens | Output per 1M tokens | Source |
|---|---|---|---|
| GPT-6 Astra (standard) | $10 | $50 | dev.to DevDay roundup |
| GPT-6 Astra Ultrafast | $60 | $300 | dev.to, SmartScope (citing OpenAI recap) |
| GPT-6.1 Sol | $2 | $10 | dev.to DevDay roundup |
| GPT-6 Luna | $0.10 | $0.50 | dev.to DevDay roundup |
Inside ChatGPT, Ultrafast draws down your subscription allowance faster. Xenospectrum, citing OpenAI's Codex "Speed" documentation, reports that Ultrafast consumes plan allowance at 8x the Standard rate (Fast mode is 2.5x), and purchased credits at 6x. Valletta Software's analysis instead assumed the 6x API ratio. Either way the math is sobering: at 8x, running Astra on Ultrafast all week turns Pro 500's 25x allowance into roughly 3x Plus worth of standard work, which is less than Pro 100's standard allowance. Ultrafast is something to switch on for the tasks where waiting costs you, not a default.
What is a dot, and which plans include one?
Dots are always-on agents powered by GPT-6 Astra, each with its own cloud computer. They connect to over 4,000 apps through plugins and work inside ChatGPT, Slack and Teams (dev.to). Every Pro tier, including Pro 100 (except for Pro users in the EEA, Switzerland and the UK at launch), and Business Premium seats include one dot at no extra cost, and chats with it do not count toward your usage limits. That makes the dot a reason to be on some Pro tier, but not a reason to pick Pro 500 over Pro 100. We cover dots in depth in our ChatGPT dots guide.
How does Pro 500 compare with Claude Max and SuperGrok Heavy?
The other two labs top out lower on price. Current list prices:
| Plan | Monthly price | Usage framing | Source |
|---|---|---|---|
| ChatGPT Pro 500 | $500 | 25x ChatGPT Plus, weekly budget, Ultrafast | OpenAI via Engadget / dev.to |
| ChatGPT Pro 200 | $200 | 10x Plus from Oct 30 | OpenAI via TNW |
| SuperGrok Heavy | $300 | Highest Grok usage limits | CloudZero |
| Claude Max 20x | $200 | 20x Claude Pro per 5-hour session, plus weekly caps | claude.com/pricing, Suprmind |
| Claude Max 5x | $100 | 5x Claude Pro per 5-hour session | claude.com/pricing |
| SuperGrok | $30 | Full model access, larger weekly pool | CloudZero |
The multiples are not comparable across vendors. "20x Claude Pro" and "10x ChatGPT Plus" are measured against different baselines and different models. What you can compare is behaviour: Claude still resets on rolling five-hour sessions with weekly caps, while ChatGPT Pro is a pure weekly budget you can spend in one sitting. If you mostly live in a terminal agent, the model matters more than the multiplier. Our Claude Opus 5.5 vs GPT-6 Astra vs Grok 4.7 comparison and Claude Code vs Codex write-up are better starting points for that decision.
Should you upgrade, stay or downgrade?
Start from how you actually use it. Check your weekly Codex and Work consumption before October 30, while you still have the 20x allowance to measure against.
Upgrade to Pro 500 if
- You regularly used more than half of the old 20x Pro 200 allowance. On October 30 that becomes more than your whole 10x budget, and Pro 500's 25x is the only single plan that covers it.
- Generation speed is the bottleneck in your day, for example long code generation, big refactors or many parallel Codex tasks where you sit waiting on output. Ultrafast is only available here and on Enterprise.
- Your time is billed at a rate where saving an hour or two a week easily covers the extra $300.
Stay on Pro 200 if
- You used less than about half of the old allowance. 10x will feel the same.
- You rely on GPT-6 Pro chat but send fewer than 100 Pro messages a week.
- You want the $2,500 credit grant. Spend it before December 31 and reassess in January with two months of real data on the new limits.
Downgrade to Pro 100 or Plus if
- You mostly wanted the Pro models and a dot, not heavy Codex volume. Pro 100 has both at 5x Plus for half the price.
- Your Codex use is bursty. Plus at $20 with purchased credits for heavy weeks can be cheaper than a Pro plan you underuse.
- Most of your heavy work runs through the API anyway. A mix of GPT-6.1 Sol at $2/$10 per million tokens and Astra only where needed may cost less than any subscription. See our Sol and Luna guide for the cheaper tiers.
For teams
Do not put a team on individual Pro plans. Business Premium seats at $100 to $125 per user include a dot and 5x Standard usage, and you can mix Premium and Standard seats in one workspace. If a team genuinely needs Ultrafast inside ChatGPT, that is an Enterprise conversation.
Timeline of the ChatGPT Pro changes
- April 2026: Pro 100 launches at $100/month with 5x Plus Codex usage.
- September 10, 2026: New sign-ups for the $200 Pro plan are paused.
- September 29, 2026: DevDay. Pro 500 launches, Pro 200 reopens with a 10x allowance, Ultrafast and dots announced.
- October 29, 2026: Last day existing Pro 200 subscribers keep 20x and 200 GPT-6 Pro messages a week.
- October 30, 2026: New Pro 200 limits apply to everyone.
- December 31, 2026: The 62,500-credit (about $2,500) grant expires.
For the models behind these plans, see our GPT-6 Astra complete guide.
FAQ
How much is ChatGPT Pro 500?
$500 per month before tax. It gives 25x the ChatGPT Plus allowance and is the only self-serve plan with GPT-6 Astra Ultrafast.
What are the new ChatGPT Pro 200 limits?
From October 30, 2026, Pro 200 includes 10x the Plus allowance for Codex and Work (down from 20x) and 100 GPT-6 Pro chat messages per week (down from 200). The price stays at $200 a month. New subscribers get these limits immediately.
Do existing Pro subscribers get compensation?
Yes. Existing $200 subscribers keep their current limits until October 29 and get a one-time grant of 62,500 usage credits, worth about $2,500, which expires on December 31, 2026.
Can I get Ultrafast on Pro 200 by buying credits?
No. In ChatGPT, Ultrafast is available only on Pro 500 and Enterprise. Buying credits on Pro 100 or Pro 200 does not unlock it. Developers can use Ultrafast through the API at $60 input and $300 output per million tokens for GPT-6 Astra.
Does Ultrafast use up my allowance faster?
Yes. Reports citing OpenAI's Codex speed documentation say Ultrafast consumes plan allowance at 8x the Standard rate, versus 2.5x for Fast mode. Running Ultrafast full time shrinks Pro 500's effective allowance to roughly 3x Plus of standard-speed work.
Is there still a five-hour limit on ChatGPT Pro?
No. ChatGPT Pro uses a weekly usage budget, and OpenAI said at DevDay that the five-hour limit will not be reintroduced.
Is ChatGPT Pro 500 more expensive than Claude Max?
Yes. Claude Max tops out at $200 a month for 20x Claude Pro usage, and SuperGrok Heavy costs $300 a month. Pro 500 is the most expensive consumer AI subscription of the three, and the only one with a dedicated speed tier.
Sources
- Engadget: OpenAI adds $500 Pro subscription, nerfs its existing $200 tier
- The Next Web: OpenAI halves Pro 200 usage and launches a $500 ChatGPT plan at DevDay
- dev.to: OpenAI DevDay 2026, every announcement with prices and availability
- OpenAI Help Center: About ChatGPT Pro tiers
- OpenAI Developer Community: Existing $200 Pro subscribers should keep their current usage allowance
- Yahoo Tech: ChatGPT's new plan costs $500 a month
- Business Today: ChatGPT now has a $500 monthly plan
- MakeUseOf: OpenAI is giving away $2,500 worth of free credits
- Xenospectrum: What "up to 8x faster" actually means
- Valletta Software: Pro 500, 25x and a $200 plan cut
- SmartScope: What is ChatGPT Pro 500?
- Thurrott: OpenAI launches $100/month ChatGPT Pro plan
- Appwrite: ChatGPT Business Premium seats pricing
- The AI Career Lab: ChatGPT Go vs Plus vs Pro (Sept 2026)
- Anthropic: Claude plans and pricing
- CloudZero: Grok pricing in 2026
If your team is building on these models, Codersera can help you hire vetted remote developers.