Message cap way below 40 for plus users?

[clear violet]

Bug Report:

Steps to reproduce:

• wait 3h to have the full 40 messages / 3h
• post some prompts until you reach the capping error message

Expected result:

• at least 40 prompts / messages every 3h

Actual result:

• i got only 20 messages every 3h, tried twice today, will try again in the evening

Additional information

Browser: Chrome
OS: Linux Mint 21.2

[astral ice]

Hello, are you using regular GPT-4 or some specific GPTs?

[clear violet]

happened with custom gpts but gpt-4 was capped then also

[tame lion]

same here
https://discord.com/channels/974519864045756446/1196439958907523082

[clear violet]

and again capped at 20 / 3h

[clear violet]

I'm having the same issue! To help, it only occurs around 13 UTC time for me or midnight here in Australia

[dull hinge]

same. I feel cheated and betrayed.

[astral ice]

An OpenAI employee recently confirmed that sometimes the cap is lowered during peak usage times to maintain stability, here's a link: #1196439958907523082 message

[dull hinge]

this is unacceptable to me; they should not call it a cap. I understand if there is a need to maintain stability, but they should use a different message. When you pay for 40 cap and you know you didn't use 40, you are right to feel defrauded, and gaslit IMO.

[astral ice]

To be fair, the usage cap being at a certain number is not part of the terms of ChatGPT Plus/Team -- access to GPT-4 and other beta features are, but the usage cap has always been variable.

[dull hinge]

mmm interesting. Where did i get the notion that it was 40?

[astral ice]

It's included in the UI in the model selector on the web version, but isn't part of the plan/payment/checkout process.

[dull hinge]

I see. I see the distinction, but I don't think it makes a difference.

My suggestion is to alter the message in the popup saying "we are slow right now" instead of "You've reached the current usage cap"

this would have saved me a lot of anxiety, anger, and time.

I bet it is doing the same to others

as it is, during this time, I am using another service 🤷

[astral ice]

I think it's a good suggestion. In the link I shared to the OpenAI employee's comment, you can see he asked for feedback like this, so they're definitely thinking about better solutions for this. You might consider adding your voice to this popular suggestions post: https://discord.com/channels/974519864045756446/1164800498096930877