I am new to Gemini API. I made a project and setup an API key for it. Now the limit of RPD has been exhausted. The documentation shows that it will get reset everyday, which should technically because its is “Requests Per DAY”. But Nothing is happening since past week. Stuck at 21/20 for past 7 days+.
can anyone enlighten me about the limits or if there is an issue.
Hello! I’m sorry to hear you’re seeing issues with rate limits not resetting daily on the Gemini API. Could you please provide more details on which model(s) is this affecting for you.
When you say “Stuck at 21/20 for past 7 days+” - is 21/20 the RPD and which model is that for? Any details are greatly appreciated as we debug this.
I have exactly the same problem and I’m using Gemini 3 Flash Preview through api key (free tier)
any updates? i encounter the same problem. i have paid tier 1 @Alisa_Fortin
Hi Alisa, thanks for taking the time to reply.
For your question, Yes the 21/20 is the RPD Limit that is being shown on the Usage and Billing>Rate Limit page.
Technical Details- Model used: gemini-2.5-flash
Accessed via API in Typescript code -
import { GoogleGenerativeAI } from “@google/generative-ai”;
the official docs say :
Rate limits are applied per project, not per API key. Requests per day (RPD) quotas reset at midnight Pacific time.
But the limit is not being reset as per this timezone too.
Other Important Details that might help:
Tested out other models, pinged them and the limits are stuck for each model.
Thanks.
Hey Aaryan! See the text under “Rate limits by model”, the rate limit dashboard by default shows you the “Peak usage per model compared to its limits over the last 90 days”, to the limit is indeed being reset every day, but unless you hit that limit again, you won’t see this number change.
Or said another way, the rate limit page is not designed to track all current usage, rather it is designed to track the rate limits and how close your peak usage got to them.
See the Usage page for more up to date daily breakdowns of usage.
Hello Logan, Thanks for the reply.
I get that it is just a visual track of Access over a fixed period. But the API calls are being rejected because of Rate Limit Error. When I switch the time range to 28 days, it says “You have reached a rate limit. Set up billing to increase your limits and unblock your work.”
And when I select 7 days, “No usage data available” is being shown. Which concludes that Rate limits are stuck and not getting refreshed on my account. I would like to add that I am sending only ONE request per session so there is no possibility of multiple requests at once. Unfortunately I don’t have my IDE logs, which had the error code, but it said something like- “Model Limit Reached” even after a full 30 hours wait.
I have used APIs of other popular LLM services, but only Google GenAI seem to have this problem. While researching for this error, many users reported it might be a bug or a “shadow ban”
Thank You!
