I have prototype RAG application built using gemini-2.0-flash for chat completion and gemini-embedding-001 for embedding generation. the API worked perfectly fine even in the last month while in free tier, but from last week it starts to throw 429 errors for the chat completion and 404 errors for embedding generation. So, I setup billing for a new project (currently on pair tier 1) and tried with that API key. Yet, even in paid tier 1 i get the same errors. even though i am getting the 429 error, as shown in the AI studio the model hasn’t hit its rate limits. 429 errors still appeard even though i change the chat model to gemini-2.5-flash.
![Screenshot 2026-02-19 at 9.41.25 AM|690x398]
P.S-: even though default output dimension size of gemini-embedding-001 is 3072, I set it to 768 because the vector db stores the vectors of that dimensionality and documentation says gemini-embedding-001 supports dynamic output dimension sizes