429 RESOURCE_EXHAUSTED but I can't see what or why?!?

Google Cloud Build with AI Generative AI & Foundational Models
1

I’m running an app coded in NodeJS and using the genAI library to make calls to either the gemma-4-31b-it or gemma-4-26b-a4b-it models.

It’s been running for months with no issues, I’m on the free tier, have no prepaid crerdits in my account and never have. All the graphs in AI Studio show that I’m miles away from hitting any rate limits, either for rpm, tpm or rpd

Suddenly two days ago I started seeing 429 errors and RESOURCE_EXHAUSTED in the logs, and all requests are failing.

I have no clue what resources is exhausted, why, and why now?!?

Any ideas gratefully received?!?

The ML request code is here: https://github.com/jodrell2000/mrRobotoV3/blob/main/src/services/machineLearningService.js

2

Hello @Jodrell,

I would say that this behavior falls under free tier usage.

From Gemini API rate limits documentation:

Specified rate limits are not guaranteed and actual capacity may vary.

When you get 429 errors and RESOURCE_EXHAUSTED, that means the Google API is overwhelmed and cannot serve your request, even if you’re not reaching your rate limit.

You have 2 options:

1/ Implement a retry with backoff mechanism, which may increase your application latency
2/ Invest some money to upgrade your tier, which should help, or use Gemini via Vertex AI instead of the Gemini API

Best of luck.

4

My thoughts as well, but the application has been running for months with no issues, and someone else running exactly the same codebase is getting requests returned.

the application is built and deployed in such a way as to make it free to anyone deploying it, so I don’t want to spend my way out of the problem…I need a way to actually see the problem :-/