I’m so glad this forum exists! I genuinely thought I was the only one with this complaint.
Charging the same $200/month while cutting our included usage in half? I honestly thought I was misreading the email at first.
I absolutely agree with everyone saying existing Pro 200 subscribers should be grandfathered into the usage allowance we originally signed up for.
And the explanation that the models are becoming more capable and efficient, therefore we need less usage, makes even less sense to me.
The models were always supposed to get better.
When OpenAI originally introduced the $200 Pro plan, it described Pro as providing scaled access to its best models and tools, said it expected to add more powerful and compute-intensive features to the plan over time, and specifically positioned Pro users as being at the cutting edge of advancements in AI.
In other words, improving models wasn’t an unexpected bonus that suddenly needs to be offset by taking something away from subscribers. Continual model improvement was part of the value proposition of paying $200/month for Pro in the first place.
If GPT-6.1 is more capable than GPT-6, that’s great. But presumably GPT-6.2 will be better than GPT-6.1, GPT-7 will be better than GPT-6, and so on. That’s technological progress—not a justification for reducing a paying customer’s included usage by 50% while continuing to charge the same price.
Especially for existing Pro 200 subscribers who signed up specifically because this was the high-usage tier.
Grandfathering existing Pro 200 subscribers at their existing allowance seems like the fairest solution.
I think that there should not be a difference in who had 200 pro or new 200 pro , would just upset new users but , it should just remain unchanged, and honestluy it shoul dbe like
!00 Usd = 10 x
200 usd = 20 x
300 usd = 30 x
400 usd = 40 x
500 usd = 50 x
if some people are making 2 accounts of 100 and 5 x base that is already giving for 300 usd the same of proposed 500 .
with the actual subscription I consume the whole amount in 3-4 days of normal work , and I am not using Astra all time but sol 6.1 , using astra just to plan or coordinate in some cases, while if they halved the amount of usage, I would be able to work only 2 days per week , whicho honestly is unacceptable .
as for the credits, I am not sure why , but they consume faster than api credits, in 1 hour I burned 1500 credits in the little time they delayed the new renewal , so I had planned to reach 0% usage at the moment they promised the resret , but the rest come late after more than a hour and I found the chat consumed 1500 credits of the 60k provided .
credits do not add value but are just a sinkhole of money.
I’m a US$200/month Pro subscriber and a hobbyist paying out of my own pocket to improve my programming skills and build my own apps. This is already a substantial expense, and I’m actively reducing unnecessary usage by choosing models and reasoning settings more carefully.
Halving the included allowance while keeping the price unchanged is an effective doubling of the price per unit of usage. Calling the models “more efficient” does not demonstrate that I can complete the same work with half the allowance.
The announced cut targets Pro 200, removing the value that justified paying for that tier. Meanwhile, moving to US$500 would mean paying 150% more than I currently do for only approximately 21–25% more allowance than the old US$200 plan. Ultrafast does not solve my problem. I need affordable capacity, not a more expensive way to consume it.
Existing subscribers should retain BOTH their current price and allowance for as long as they remain subscribed. Introduce different terms for new customers and offer premium tiers to those who want them, but preserve the deal for people who have already built their working routines around it.
One month’s grace and temporary credits are not substitutes for a recurring allowance. My projects and learning do not end when those credits expire.
Against reports of trillion-dollar valuations and a possible IPO, cutting existing subscribers’ allowances in half is particularly galling. If OpenAI wants investors to see a sustainable business, it should demonstrate that customers willingly pay for a dependable service. Reducing what loyal customers receive while pushing a substantially more expensive tier damages that trust.
Please preserve existing subscribers’ allowances and publish clear, comparable limits for every tier. At this price, customers should not need independent researchers to work out what their subscription actually buys.
I’m seriously considering moving to Anthropic. For people using Claude Code heavily: does its allowance hold up in practice, or are you encountering similar reductions? I want to spend my time learning and building, not repeatedly reorganising my work around a shrinking subscription.
It even worse than expected here. I actually upgraded to the $500 per month plan from my $200 a month plan because I wanted to utilize Ultrafast.
So just before my subscription change on my grandfathered 200 plan. I ran 6.1 Sol in Fastmode for 18.5hrs and burned 15% of my weekly limit. After upgrading to the $500 plan in the same session 6.1 Sol Fastmode (not ultra) I’ve burned 4% of weekly in just 2.5hrs. This is a goal btw so it’s continuous operation refactoring and rewriting about 1500 markdown documents with humanizer. The documents don’t magically get harder through the set they’re all very consistent because they’re plan documents that followed a template during initial design. That said let’s extrapolate:
So.. 20x to 25x is a complete lambasted lie straight through their teeth or I’ve missed some documentation that says otherwise. This may be anecdotal for right now so I’m going to start comparing token for token over the weekend. Bottom line, we’ve all been bent over boys and girls. Just sayin.