Please remove the Plus limit for your paying customers!

I am more than happy to pay for good programs and I don’t agree with people who think everything should be free. So after using GPT3.5 for a couple of times, I upgraded to plus for 23€. Imagine how surprised I was when I found out the hard way that there is a limit of a few dozen messages (which I didn’t even use) and that I should come back later.

Sorry if I sound harsh now but how is that a fair plan? The normal way should be that the FREE 3.5 version is limited while you pay to get unlimited prompts. If you are low on server capacity then cap the free version or make them pay 5 $ (which many would do), not annoy those who are willing to pay with such business models.

In addition to that, there are several other problems I encountered:

  • every prompt produced consistently 1-3 error messages, sometimes more. I had to regenerate the answers which apparently counts against my limit. In some cases I had to rewrite the question because it produced errors over and over again.
  • I couldn’t subscribe to the plus version with my browser. I had to google to find out that this is a bug and subscription only works with the phone app.
  • After you get a message “you have reached the limit” you can click on a link that tells you more about this limit no one has heard about before he reaches it. This link is broken and leads to an empty page.

I am okay with the bugs but the limit policy for paying users is a no go. I am going to cancel the plus subscription again until this has been changed. GPT is an awesome piece of software I didn’t expect before 2030 so it is sad that there is no plan to get unlimited access.

2 Likes

I think you’re confused, non-paying users have zero access to GPT-4. Paying users have limited access to GPT-4 and (effectively) unlimited access to GPT 3.5.

I didn’t confuse that. I just think that unlimited access to 3.5 for free vs limited access to 4.0 for 23€ is bad. It should be 3.5 costs a low amount of money or is capped while paying means you get unlimited access to 3.5 and 4.0.

1 Like

I don’t think you understand how much it costs to process a message using the GPT-4 model.

You do still get (more) unlimited 3.5 use as a Plus subscriber, the real value though is in the GPT-4 access.

If you can’t extract (in your case) 23€ of value from (potentially) 9,600 interactions with GPT-4 I believe either,

  1. You simply don’t have a good use case in your life which requires (or benefits from) any of the unique capabilities of GPT-4, or
  2. You need to learn a bit more about using and interacting with LLMs to understand better how to get better outputs more efficiently and reliably

That’s not a knock on you, there’s a learning curve and every use case is unique in the challenges to overcome in getting the models to excel.

It could be that 3.5 is good enough for what you’re doing, in which case that’s fine, Plus might just not be right for you, which is great for you because you can get all you need free.

If it’s as expensive as you say, then why not add another tier - let’s say 30 or 40€ - to gain unlimited access. And price the 3.5 model at 4,99€ a month. Even a pupil could pay that for a month of free homework.

My problem is - asides from the bugs that force me to regenerate 70% of my messages and swallow all my limits - that I have certain times when I need GPT intensively, and then I have long downtimes where I don’t need it at all. At least offer different types of subscription plans for different kinds of users.

They released “Teams”… Which. I’m not really sure if it’s an improvement. I’d actually consider it (in regards to message limit capacity) a downgrade.

100 messages / 3 hours PER TEAM. Maybe this is incremental based on the team count. Also not sure if it follows the same structure with Custom GPTs being half. If you’re desperate or know someone who barely uses it you could buddy up and take advantage.

Realistically, I think these limits are reflecting how much they can actually output. I don’t think it’s necessarily about money. The hardware required to generate these tokens are hard to get a hold of.

Yeah. There is clearly a spike in network issues happening right now. Almost guaranteed to happen when some new feature is released. It’s also frustrating that we can’t “bank” these messages, or even roughly know how close we are to hitting it.

It was only a couple weeks ago near the release of CustomGPTs that purchasing ChatGPT premium was temporarily blocked and this tightened limit was introduced.

I just found out that bing uses GPT4 aswell and is free, apparently without limits.

Yes, I do acknowledge that they are not necessarily adept at conducting business.

However, it’s important to understand the specific limitations of Microsoft’s Bing Chat, which is designed primarily for search and is subject to a cap of 30 dialogues per session due to strategic and resource considerations.

Microsoft’s significant investment in OpenAI has allowed them to implement features that may not be available to others, but this doesn’t mean there aren’t alternatives for those with the technical expertise and resources.

GPT-4 is a state-of-the-art technology that requires substantial computational power, and even with Microsoft’s extensive infrastructure, it’s not feasible to offer unlimited access without some constraints.

Unlimited access to GPT-4 would require a dramatic increase in computing power.

In the meantime, those with the necessary skills and financial resources can develop their own services within their means, which could potentially fill the gaps in the market.

If you’re in search of services tailored to your specific needs, it may be worthwhile to look for individuals or companies that specialize in this area.

While I can’t provide direct business recommendations, a careful search might lead you to a provider that can meet your requirements.

1 Like

Alright, what I got from this chat is that with our current technological level, it is too cost, energy and hardware intensive to grant unlimited access. I would still like to see a tier above the current plan but maybe I just have to wait. Who knows what we have in a couple of years. Thanks for all the replies.

2 Likes

There just isn’t enough capacity to allow unlimited use. There aren’t enough GPUs with AI companies all vying for them. Bing has unlimited use because they, if I’m not mistaken, are running it on their own infrastructure in Azure, but someone can’t correct me if I am wrong. I’m not sure it’s unlimited. There may be limits.

Microsoft is providing its cloud service, Azure, to OpenAI with a higher priority than Microsoft itself.
As a result, Microsoft is in a position of using Oracle Cloud as a customer.
Please refer to this article for more information.

I believe you can also read the following with Google Translate.

I agree with you, how surprised I was when I just paid $20 to upgrade and a couple of hours later I get the “you’ve reached the current usage cap” like WTF did I just pay $20 for it will be the first and last $20 they get from me thats for sure.

2 Likes

I understand that asking questions to ChatGPT costs resources and money.
So why not pay per token as used in the API.

Or pay something extra if you are running out of your “bundle”
My issue is that I have a paid version because I use it professionally. Being blocked for a few hours costs so many more.

My biggest issue is that OpenAI does not provide a way around it. Even not paid.

i have paid before 4 days to upgrade to plus version. today i opened the browser and got session expired, then i tried to log in , only the version 3.5 is up there, and if i want to use the plus it shows me that i have to pay. i don’t think that is true cuz it’s only 4 days passed. any recommendations???

I too have just hit the limit of chats and feel very cheated when other AI services offer limitless usage for free (Gemini, Pi). I do feel cheated because this is not made clear when paying the $24 per month. This is particularly frustrating if you are trying to refine a GPT you are creating and need to experiment multiple times to refine outputs. Or the outputs that are given are too general, and despite specifying for further detail, require additional attempts to get to the specifics.

People are jobless pretty sure they cant afford it. They are using AI to break from their jobs also. Catch 22

I think if they want to cap paying users. Why are we paying it for it. There is no benefit at that cap. We cant use it to do it anything of value. I dont think anyone could. We’re better off cutting them at the knees all at once. By stop paying on mass. That should get the ball going. Complaining does nothing. They need to see we mean business.