I’m also getting this with VertexAI. It seemed to start yesterday and now is erroring on practically every request.
I have also the same problem. I am on pay per use, under quota with Vertex AI and still getting random 429 resource exhausted
@Sonali_Kumari1 or anyone from the Google team,
I’m also often getting this 429 rate limit, while not hitting my rate limits.
- Using Gemini 3.0 Pro in Cursor (via my Vertex AI API key)
- Project Number: 900009280905
- Free Plan with Startup Credits (Tier 0)
- Can you fix that so that I don’t get these rate limits?
- What is actually the limits on Tier 0? What is Tier 0? Can you move me up if relevant (I’m on a higher GCP startup program package)
IMPORTANT
3. I need to understand how reliable or unreliable Vertex AI will be via my GCP account, as I’m developing a Voice AI app and was planning to use Gemini Flash 3.0 to power my app.
Thanks
Been having the same issue on my project since Dec 2205, using Gemini through Clay where API is the only authentication method.
Pro plan…
agent executor error: RESOURCE_EXHAUSTED (code 429): You have exhausted your capacity on this model. Your quota will reset after 118h52m56s.: RESOURCE_EXHAUSTED (code 429): You have exhausted your capacity on this model. Your quota will reset after 118h52m56s.
This is frustrating as hell. The model either returns 429 way too often to be visible. Google keeps this model in preview while providing a horrible service for paying users. What’s going on Google?
This is still a problem for me. I couldn’t figure out how to DM you the project number
- it’s been 2-3 months and now not even your partners trust you to solve this issue. Clay is even deprecating this option because it’s unstable, causing a lot of trouble for the end users.
This is what happens when the motivation for making money exceeds the motivation to make a good product.
Any luck solving this? badly stuck with it and at standstill. paid billing account changed model asked gemini too, nothing is of any use…
Gemini Error fetching live data:
got status: 429 RESOURCE_EXHAUSTED. {“error”:{“code”:429,“message”:"You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limit
same here. this shouldn’t happen for longer than an hour. google gemini api is unusable for me for the past few days. embarassing …
I used to get this all the time so would switch up models to avoid once 429 hit.
Nowhere near any VISIBLE rate limits, but I think it averages out and blocks based on what it ‘thinks’ that usage will come out to over time. I would ‘blitz’ the api with thousand of requests and then not use if for days. Just the nature of the work.
I’ve switched over the batch requests now which were working fine, but even that has gone bad in the last 3-4 days.
I have 5 jobs coming up on 24 hours ‘pending’ which is ‘technically’ inside service level, but before the last few days it would be 1-4 horus.
Hi,
I do use Batch API with Tier-2 limits, although I’m nowhere near the limits Gemini just doesn’t allow me to create new batches with the same error, RESOURCE_EXHAUSTED. So this issue still persists.
I started getting this error in the last few days. I started to switch models to avoid it and now it just blocks. I use Gemini in Vertex AI Studio and get this error. However when using via the API it seems fine? I have searched all of our quotas none are even close to max. THe way I use gemini is bursts of high activity ~100 quick requests inside of a few seconds then nothing for hours. We are a paid enterprise account and the project has sufficient billing it shouldn’t be ANYWHERE close to a rate limit.
What is going on? This is a recent issue for us so something obv. changed. I also started to see a pop up ‘
Error message: “Cannot read properties of null (reading ‘candidates’)”
‘
on the Vertex AI Studio
Still no solution? I use Tier 1 with 3.1 Pro in Cursor and I get User API key rate limit exceeded since yesterday around 19 hours (GMT+1), so I’m already way past midnight PT
Plus one. I did not hit the limit but the API keeps blocking me. Google please fix this is significantly painful and hurt our business
Hello, it’s a rate limit problem. Just switch to another model within the project inside Google AI Studio. If you are using 3.1 Pro Preview, go back to 3 Flash Preview. It will work for you if you explain it what to do properly. Don’t oversaturate the requests and ask the model what is a healthy amount of information. It will work wonders for you while 3.1 Pro is being adjusted and worked on.
Was anyone able to solve this?