Persistent 429 "Resource Exhausted" on gemini-embedding-001 even with fresh keys/projects

Gemini API
, , ,
1

Hi everyone,

I am building a RAG pipeline and using models/gemini-embedding-001 to process a relatively small dataset (approx. 65 markdown files).

The Issue: I am encountering persistent 429 Resource Exhausted errors that seem unrelated to my actual usage volume.

  • I successfully processed a small batch of files, but suddenly hit a wall.

  • Currently, I get an immediate 429 error on the very first request, even after waiting for hours.

  • The retry-after header returns spiraling wait times (e.g., 320s+), and the request never succeeds.

Troubleshooting Steps Taken:

  1. I implemented exponential backoff in my code, but it doesn’t help since the initial request is blocked immediately.

  2. I tried creating a fresh API Key on a new project to rule out quota issues, but I still receive the same 429 error instantly.

  3. I noticed other users reporting similar “phantom” 429 errors recently (e.g., Link 1, Link 2).

My Questions:

  1. Is there a known issue or degradation with the gemini-embedding-001 model specifically?

  2. I am planning to upgrade to the Pay-As-You-Go (Blaze) plan to resolve this. However, seeing other threads where Paid Tier users are also facing 429 errors, I am hesitant. Will upgrading guarantee a fix for this specific error?

Any advice is appreciated.

Thanks.

2

Welcome to the club.

1 Like
3

I am on gemini 3 pro (no images), on AI studio
Was using it happily with the API key
Suddenly hit quota exhaustion errors too, despite very low usage rates.
This has never happened before while on quota. I suspect there must be a bug somewhere

1 Like
4

Same problem with you

6

Hi all,

We truly appreciate you flagging this issue and apologize for the inconvenience. Could you please provide the project number (not the project ID) via direct message?