Consistent 429 from Gemini API despite being within quota

We hit TPM quota on Tier 1 for a few minutes and started getting 429s from the Gemini API.

We stopped the traffic for over an hour now and continue getting 429s despite having nearly 0 input tokens sent for quite some time now.

Is hitting a TPM quota at any time putting account into essentially unusable state for an undefined period of time?
What are the cooldown periods?
How to get out of this bad stage now which seems to be undocumented?

Same issue for me, I dont understand how these big AI companies dont have this issue

1 Like

same problem. hope resolve asap….

Thank you for reporting, will check with the team.

Thanks. this problem is “TPMinute –> TPHour”, which means the api token limit didn’t refresh per minute as the rules but per hour.

Same issue here, mega super annoying :frowning:

2 Likes