Gemini API Errors

Are any of you experiencing 429 errors this morning? I’m seeing 35% of requests return quota errors (429), even though we haven’t made any code changes in production or experienced any spikes (yes, we have a paid api account; I also tested with 2 requests over 60 seconds).

This behavior is similar to the 503 errors we encountered a few weeks ago, which we resolved by implementing a backoff solution. The backoff worked well for over a week, but now this issue has surfaced.

Please let me know if you’re encountering similar errors.

Thanks,
Damian

5 Likes

Same here, I noticed the issue this morning. And never experienced it before. I can also easily reproduce by sending consecutive requests containing an image.

3 Likes

I’ve been having 503 errors, not 429 errors (although I have had weird 429 issues before)

Are you using context caching? Whenever this has happened to me it’s been when using caching

1 Like

No, I’m not using context caching

1 Like

I’ve been getting 429 error a bunch lately, too.

1 Like

Hi all,

Please follow the below instructions to troubleshoot:

Go to GCP console and click “APIs & Services”. Under Metric, search and select “Generative Language API”.. Under “Quotas & System Limits” tab, check for “Current Usage percentage”..

If it reaches 100%, then you have reached your quota limits.

If you think that there is any discrepancy, please DM me with a clear error message and Project ID to help us investigate further.

1 Like

gen-lang-client-0795798814

I get it with Gemini 2.5 pro and flash while I have this $300 credit thing. I get one or two messages then it’s spams me with this: Google AI returned error code: 429 - Resource has been exhausted (e.g. check quota). Maybe because I send or request multiple times in 2 minutes?

@Krish_Varnakavi1 see above

Got it.. I am escalating your issue and will ask Engineering team to investigate.. In the meanwhile as you mentioned that you requested multiple times in short window, please try again and let me know if you are still facing the issue.

Thanks for your patience.

1 Like

@Krish_Varnakavi1 it seems to be fine after a few minutes but doesn’t like it if I send a request after a turnaround of <1min

Due to imbalance of active users to server scaling, models response sometimes may encounter such issues.. If you have a use-case that needs to be productionized, please check my other post where I helped a user implement retries with exponential back-off for their code…