Monthly set payment limit being exceeded


I am on a pay as you go account for api usage. I am setting my monthly budget via the Settings → Limits screen on my account to $1 and expect to get errors returned by the gpt api once the budget exceeds $1. At the moment I have already spent $1.32 and am still getting 200 responses from the api. Why is the usage limit I am setting not being applied?

Welcome to the dev forum @elektra

The usage limits page says that there may be a delay in enforcing the limits:

Usage limits

Manage your API spend by configuring monthly spend limits. Notification emails will be sent to members of your organization with the “Owner” role. Note that there may be a delay in enforcing limits, and you are still responsible for any overage incurred.

Usage limit

The maximum usage OpenAI allows for your organization each month. View current usage

Set a monthly budget

If your organization exceeds this budget in a given calendar month (UTC), subsequent API requests will be rejected.

Set an email notification threshold

If your organization exceeds this threshold in a given calendar month (UTC), an email notification will be sent.

1 Like

Yes thank you, I read this but that is too vague. I implemented this change before the weekend and still hasn’t been applied. The email notifications do come through already. Might you have any experience with what timeframes we are looking at regarding the “delay”. I initially had imagined a few hours would be reasonable, a day at most. 3 days seems a bit long to me

OpenAI had previously tried out completely hiding this monthly limit option from pre-paying users. Stupidly, as then for those that had set a limit, it still was working, and couldn’t be changed. That was shortly fixed after a day of cutting off developers.

And then after a week of screaming, they put it back.

Could it be that the option is still made useless to stop usage over the setting?

The expected delay would likely be similar to the propagation delay in an API being disabled after use - sometimes minutes, sometimes much longer.

The largest fault in any delay at all is in outsourcing not just the payment collections, but the very accounting of every API call, to CEO buddy buddy Stripe.

Since there is no OpenAI advocate actively engaging here, the place would be API->Feedback in the help option of the API platform account.

Thanks for getting back.

Unfortunately (for this situation) I have never had exceeded my monthly usage limit.

I remember there were some errors where once the limit was set, it couldn’t be changed, so I let it set on default.