To create any sort of Custom GPT with this usage limit is pretty much impossible. It would take years, lol. Imagine working on it, then after 20-50 messages you get cut off for 3 hours. Yeah that kills the creativity, inspiration and joy all at once.
20 USD for 40 messages per 3 hour = Cancel my subscription after 1 month.
GPT 4 was never advertised or offered as an unlimited service. You pay $20 for effectively unlimited access to GPT-3.5, limited access to GPT-4 including GPT-4V, web browsing, and the advanced data analysis tool, as well as the ability to create and use custom GPTs with information retrieval and plugins.
If you are unable to extract $20 worth of value each month from this, it may be that you just donât yet have a good use case.
The good news is, over time, the models will get more capable and cheaper to run so sometime soon you should hopefully find a value proposition more to your liking.
I still agree with @elmstedt, it may seem difficult to feel like youâre getting your $20 worth at first, but as you learn how to use it the value it can add increases dramatically.
I know that sounds like a lot of unnecessary work and you can get what you need from bard, and if thatâs true then go for it. GPT Plus definitely can add much more value than it costs if you use it well though.
P.S. The advertising material does mention that enterprise is the only plan with unlimited GPT-4 access.
I can extract a lot of worth for $20. I know how to use this platform and itâs very helpful, but how the h3ll Iâm gonna be productive if I need to continue my work and this limit interrupts me?
Imagine paying a platform to be productive and still have to wait 3 hrs to continue my work,
And I was only in my 19th input, not 40 as mentioned.
I absolutely understand your frustration. A lot of us have API accounts anyways and have started using those instead for what youâd use ChatGPT for. Theyâre a little more expensive overall, a little clunkier, but you donât get these service interruptions.
Getting usage capped right when youâre in the middle of your flow is the absolute worst. 
I understand itâs not ideal, but it is what it is. On the Plus plan you get about one message every five minutes.
There are ways to mitigate it though.
One thing I do is lean more heavily on 3.5 than 4âat least initially. 3.5 is also quite capable so thereâs often no need to use 4 for something 3.5 can do. By directing requests at both models the 40-messages/3-hours is spread out more.
Another thing you can do is to not rely on the model for everything. Presumably you did what you are doing now before you had access to GPT-4, so instead of offloading everything to the model, use it more judiciously.
As previously mentioned by @Diet, you can use the API to query any of the models in a pay-as-you-go fashion.
If youâre working with others, you might consider a ChatGPT Teams account as those have a higher message limit of 100-messages / 3-hours or about one message every two minutes.
At the end of the day the fact is the heaviest users could easily burn through tens or hundreds of dollars a day of GPU compute if they had unlimited access to GPT-4, so they canât do that. Even at 40-messages / 3-hours thatâs 120-messages in an 8-hour workday.
120-messages could easily be a half-million input tokens and 250-thousand output tokens. Which is the equivalent of $12.50 in API calls.
Even if youâre just using it for âworkâ thatâs $62.50 / week or $250 / month in equivalent API calls (not even including any additional tokens for system messages).
In theory you could potentially use ~9,700 messages a month with ~4,000 input and ~2,000 output tokens each.
At $0.01 / 1K input and $0.03 / 1K output thatâs ~$0.10 / message. So, if you really tried to push the limits, you could probably get the equivalent of $1,000 worth of API calls out of your $20 / month subscription.
So, while I get itâbeing interrupted by hitting the limit sucksâyou may want to moderate your expectations a little bit. A year ago you couldnât get any access to anything even approaching the capabilities of GPT-4 for any amount of money you wanted to throw at it. Now, for about $0.66 / day, you get access to the most powerful language model ever publicly released.
Iâve been in these situations, too, and as all of us in this post agree that it can be frustrating. But my take-away was that I need to work on my time management and just better plan my workload around these constraints if and when I use ChatGPT to complete certain tasks.
If thatâs not possible based on the nature of your work, then the API is ready and available. Mine runs 24/7 but that âluxuryâ comes at a higher cost. But compared to what it would cost to have the same tasks to be performed by a human, itâs still nothing.
It is not about being less than ideal, it is the fact that it is not viable. Plus with its features and functions at 20 is viable without cap, but not with a cap. Increase the price OpenAI so this becomes useful. Right now it is a joke. 40 msg per 3 hour is ridiculous.
Teams has a limit of 100 / 3 hours and Enterprise is truly unlimited.
If neither of those work for you there is always the option of rolling your own with the API.
I typically use 3.5 due to the message limit on 4 and the fact that 3.5 is incredibly fast. I only resort to using 4 if 3.5 canât accomplish what I need, or if I am generating images.
100/3 is still not viable. 250/3 would be decent. 200/3 at bare minimum. Paying for Enterprise is beyond what I can. I hope this gets resolved⌠because GPT4, CustomGPT and Plugins are pretty awesome, however there is no point in paying for an awesome product you cant use.
Which is hardly the point of what should happen when one pays for GPT4âŚ
OpenAI has adjusted the usage cap numerous times. Hopefully it will be increased again soon 
. They have been releasing a lot of new stuff that ultimately costs more resources
I remember using GPT-4 back when the cap way 25 per 3 hours. I thought very carefully about each question lol.
Good for us that people were willing ta patron this product when it was useless so that OpenAI could develop into something that is (hopefully) becoming useful in the very near future.
It definitely will be useful, to many ( including myself) it already is. I use it every single day, throughout the day.
Itâs still a very early-stage technology of course and has its flaws (message cap, unable to remember previous conversations, speed, etc.) but itâs important to remember that weâre getting to test tools that will likely end up changing various areas of our lives.
Iâm very young and idealistic, but I can say with 100% confidence that this tech gets me more excited than anything else Iâve ever seen.
Theyâre fascinating to learn about, I highly recommend The AI Explained youtube channel if youâre interested in learning more about it.
Tell me more about how you use it throughout the day with this enormous usage limit?
Iâm very young and idealistic
Gee⌠you dont say? I think everybody understand that when reading your reply.
This new limit is a Joke and really kills productivity
So I signed up for $20 a month to use 4. 3 days later I am hit with a âCAPâ?? Iâm paying for a CAP? Im literally paying to NOT be able to use something I just paid for? I saw nothing about CAPs when I signed up. In fact I thought I was avoiding the limits on 3.5!! Now Im paying to not be able to use 4??? Crazy and unacceptable.
Then I read the caps are to stop bots? Really? Like you dont know when a real user is signed in or cant sent a test message to confirm?
Then you check the website and can you get to contact anyone? Not a chance!
Who to complain to? FTC? Câmon OpenAi, you can do better than this!!!
