Sign in Subscribe
news

2026-09-29 ChatGPT Pro 500 - the $500 plan that came with a Pro 200 cut

Sebastien Dubois

Sebastien Dubois

— 4 min read

Canonical version: 2026-09-29 ChatGPT Pro 500 - the $500 plan that came with a Pro 200 cut.

On 29 September 2026, OpenAI added a third Pro tier to ChatGPT: ChatGPT Pro 500. It costs $500 a month, has the highest included usage of the Pro plans, and is the only Pro plan with GPT-6 Astra Ultrafast in ChatGPT Work and Codex.

The same day, Pro 200 reopened to new subscribers with a smaller allowance, at the same $200 price.

What OpenAI announced

  • Three Pro plans: Pro 100 ($100), Pro 200 ($200) and Pro 500 ($500). Monthly billing only. Pro 500 signups are web-only at launch
  • Ultrafast on Pro 500 only. You pick it in the model picker. Per OpenAI's docs, it generates tokens up to 8x faster than GPT-6 Astra in Standard mode, and it eats your included allowance at 8x the Standard rate (purchased credits: 6x). Buying credits on Pro 100 or Pro 200 doesn't unlock it
  • Pro 200 is back, with a lower allowance. OpenAI's wording: the cut is meant "to reflect our increasingly efficient models". The price stays at $200
  • Grandfathering, for one month. If you had an active Pro 200 subscription between 22 September and 10 a.m. Pacific on 29 September, you keep the old allowance until 29 October 2026. After that, you drop to the new one. Keeping the old allowance doesn't give you Ultrafast

The rest of the Pro bundle is shared across the three tiers: Astra-powered Pro reasoning, Dot (OpenAI's always-on agent, rolling out), longer Codex and Work sessions with no five-hour limit, and bigger context windows than Plus (400K vs 256K for reasoning models).

How much usage, exactly?

OpenAI doesn't publish usage numbers. The pricing page says "more usage". The help article says Pro 500 "offers the highest included usage of the three plans".

The reported numbers come from other sources. An OpenAI email to Pro 200 subscribers, quoted on Hacker News, says that starting 30 October 2026, Work and Codex usage drops "from 20 times to 10 times the ChatGPT Plus allowance", and the GPT-6 Pro chat limit goes from 200 to 100 messages per week. Commenters and the MindStudio write-up put Pro 500 at 25x Plus. Reported usage by plan:

Plan Price Reported usage vs Plus
Plus $20 1x
Pro 100 $100 5x
Pro 200 $200 10x (was 20x)
Pro 500 $500 25x, plus Ultrafast

The old Pro 200 gave you 20x for $200; Pro 500 gives you 25x for $500. Usage scales linearly with price, and the volume discount at the top is gone.

According to another email quoted on HN, existing subscribers also got a one-time grant of 62,500 usage credits (a "$2,500 USD value") that expires on 31 December 2026.

OpenAI's explanation

Tibo (Thibault Sottiaux, from the Codex team) posted the rationale on X the night before. He was upfront about the cut: "if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan". His arguments:

  • The five-hour limit stays gone, so you can spend your weekly usage whenever you want
  • Usage is now pegged to API prices, and as models get cheaper on the API, you get more work done per dollar on the subscription too. He pointed to GPT-6 Sol and GPT-6 Luna launching at half their previous price that week
  • OpenAI doesn't want an incentive to inflate API list prices to make subscriptions look generous. Long term, he expects the gap between subscription value and API value to close, so most people "buy usage as needed"

That last point implies the subsidized flat-rate deal is meant to fade out.

How the community reacted

The Hacker News thread (218 points, 266 comments) was overwhelmingly negative. The complaints targeted the Pro 200 cut and the hidden limits rather than the $500 price.

  • The Pro 200 cut. One commenter said they'd lose half the value of Pro 200 and then "have to pay 2.5x as much to get 1.25x the value I used to get". The "increasingly efficient models" justification got mocked a lot
  • Hidden limits. Several people checked the pricing page, the help article and the signed-in upgrade screen and couldn't find a single usage number. One asked whether "hand-wavey usage limits on a 500$ subscription" is crazy
  • The naming. Naming a plan after its price got jokes ("For a second I was worried I somehow missed 490 versions"). One commenter pointed out that tying the name to the price lets OpenAI quietly shrink what's inside
  • Switching talk. Plenty of commenters said they'd cancel or downgrade, move back to Anthropic (with Claude Opus 5.5 cited as the reason), or go to open-weight models like Kimi K3 and GLM 5.3
  • Some defended it. People running multiple $200 accounts can now consolidate into one. Others pointed out that subscriptions are still cheaper than API pricing, that the 8x speed matters to builders, and that newer models may be cheaper per task, so a smaller multiplier isn't automatically a worse deal

My take

The plan name states the price, but OpenAI doesn't publish the allowance. I don't want to pay for capacity I can't measure, at $20 or at $500. The pattern also fits what I described in AI API prices are rising: the frontier tier holds or raises its price, and subsidized volume is ending. If your workflow depends on a single subscription, now is a good time to know your cost per task and to keep a second provider (or the API, via OpenRouter) ready.

Ultrafast might be worth it for live, interactive coding sessions where you sit and wait. For long autonomous runs, speed matters less than how far the allowance stretches, and Ultrafast drains it 8x faster.

More on the plan itself: ChatGPT Pro 500.

That's it for today! ✨

References

About Sébastien

Ready to get to the next level?

Found this valuable? Share it with someone who needs it.

Join 6,000+ readers. Get practical systems for knowledge & AI. Free.

Subscribe ✨

Free: Knowledge System Checklist

A clear roadmap to building your own knowledge system. Subscribe and get it straight to your inbox.

6,000+ readers. No spam. Unsubscribe anytime.

Subscribe