Hi,
I just started with my first GPT.
I bought GPT-Plus and made sure I had some money on my balance on platform.openai.com.
But while developing and testing my GPT I get the error: You’ve reached the current usage cap for GPT-4, please try again after 5:15 PM.
My Question:
-
Is this by intent or does this maybe mean that something with the billing gone wrong?
-
In case there is still a limit on GPT4 - how can I switch my GPT to another version without limits? While creating the GPT I wasn’t asked to choose between v4 or v3.5.
Thanks!
Hi - Regarding you first question: For creating a GPT, you only have to be a GPT Plus Subscriber. Your account balance, which may use for using the API, is independent from that. So the good news is that nothing has gone wrong with your billing
There is indeed still a usage cap on GPT-4, both in the regulator ChatGPT interface as well as for custom GPTs. I think for ChatGPT its at 40 messages every 3h and for custom GPTs its a bit lower, I believe at 24 messages every 3h.
Hi,
Thanks for your answer!
If there is a limit, how can I test my GPT? After a few messages I will be blocked and can’t continue developing the GPT?
Thanks for your help!
I understand the frustration but everyone is equally affected by this. It’s still a new technology that requires a lot of resources to operate and this is so far the best way to manage the resource load while keeping it fair across users.
If you want to be free of constraints in terms of usage cap, then consider building your solutions with the API.
The real option to break this limitation is to go to the direct management mode of the api, you buy a token for development, and if that token is used in a response, you have to buy more tokens.
At this point gpt Builider is a tool for non-programmers, however to give some particularity to your chat bot, you will have to implement your own api using YamL or Json format making calls like get, post, put, delete etc to your own API server.
Now, in development limit the use of the chat bot unless necessary, and in local tests as api actions consider using an external server as tests, the editor https://editor.swagger.io/, is usually very useful, or in this case xamp so you can simulate a local server, that way you will only use the bot in tests that really deserve it.