Automatic token recharge not working

I woke up to a critical issue with the API today; auto recharge did not trigger (although its enabled!), the credits went to $0, and our production app stopped working for 2 hours while we were asleep… this is the second time this happened for us – what is going wrong there?

We have two credit cards setup in payment methods, and we did not receive any email that charging us failed.

I just got bitten by this. My credit went to zero and no auto-recharge happened.

On the billing page it says “Auto recharge is off”, so I assume it’s my fault and turn it back on. Refresh the page and it says “Auto recharge is on”, but hasn’t recharged. And about 5 minutes later, I refresh again and we’re back to “Auto recharge is off”!

I did a manual top up with the same credit card and that worked. So I tried turning recharge on again now that I had positive balance.

Refresh 5 minutes later and it’s still on. So maybe it only fails when a user has a negative balance?

This is really worrying! I now need to nervously check my credit levels manually for fear of auto-top-up turning off.

We experienced the same issue, and this is a serious problem for production use.

Auto recharge was enabled for our organization and appeared to be configured correctly. However, it did not trigger as expected. Our credit balance went negative, and our production service was affected.

The most concerning part is that there was no visible warning or failure indication:

  • We did not receive any notification that auto recharge had failed.

  • Billing history did not show any failed payment attempt.

  • There was no indication that the monthly recharge limit had been reached.

  • From the dashboard, everything looked normal until the service was impacted.

This makes auto recharge unreliable as a safeguard for production systems. If auto recharge can silently fail without any notification, failed payment record, or clear dashboard warning, users have no way to detect the issue before API access is affected.

Could the OpenAI team please investigate this urgently and clarify:

  • whether an auto recharge attempt was actually made

  • if it was not made, why it was skipped

  • if it was made, why no failed payment record appeared in Billing history

  • whether this was caused by the monthly recharge limit, payment processor behavior, or an internal billing issue

  • what alerting or monitoring OpenAI recommends to prevent this from happening again

This caused real production impact for us. We need a reliable way to know when auto recharge fails before our services are affected.

Been having the same issue here for the past few days. Im having to monitor manually and recharge myself to avoid production going down.

Hey everyone, Apologies for this inconvenience. We are happy to look into this with urgency, we will require some personal details which we cannot attain here in public forum. Can anyone open a support request with us at support@openai.com and share account details and exact time when you observed this behaviour and share the case id/case subject with us here and we can track/resolve it for you. Thank you!

Due to no response for several days, we are closing this thread.