Receiving quota error when trying to use bison chat model in Vertex AI

[dashy]

Hi, I want to try out the new bison chat model. However, when I'm asking anything I'm receiving this error: 

Quota exceeded for aiplatform.googleapis.com/online_prediction_requests_per_base_model with base model: chat-bison. Please submit a quota increase request.

[oiwejdsd]

Me too.

Submitting a prompt on https://console.cloud.google.com/vertex-ai/generative/language/create/text results in the following error

Failed to submit prompt

Error message: "Quota exceeded for aiplatform.googleapis.com/online_prediction_requests_per_base_model with base model: text-bison. Please submit a quota increase request."

Status: 429 Error code: 429

Tracking number: xxxxxxxxxxxxx

I've tried to follow the docs on Quotas and limits but there seem to be no quotas for Generative AI.

I'm on the GCP Free Trial if that is relevant. Unfortunately, this means I can't contact (paid) support.

[mchrestkha1]

@oiwejdsd can you try again as well and confirm that you can now access 30 queries per minute?  Make sure you're using us-central1 region.

[oiwejdsd]

@mchrestkha1 Thanks for your replies to this thread. My quota screen looks like @dashy's and is still at zero.

[mchrestkha1]

Here are the quotas https://cloud.google.com/vertex-ai/docs/quotas.  When requesting a quota increase look for this metric in the filter 'aiplatform.googleapis.com/online_prediction_requests_per_base_model' .  You should see quotas by region and model with default values.   

[dashy]

Yes, but all quotas a set to 0 and I can't request increase quota either. It mentions that I need to contact sales.

[mchrestkha1]

could you share a screenshot?

[j_molina]

Hey! I am seeing the same issue 😔 

[mchrestkha1]

@j_molina can you try again and let us know if you can now see 30 queries per minute for us-central1 for the bison models.  Your screenshot is showing AutoML services which are different.

[dashy]

Here's a screenshot

[kolban]

Here is what my quotas look like in my paid account

Since this is a preview service and Google is currently not charging to leverage this service, it may possibly be that Google has disabled this service for free tier accounts.

[dashy]

Hi Kolban, I'm using an enterprise account (no free trial). However, I just tried with my personal account and that have quotas and I can use the models. Not sure. Thanks for checking it out!  

[kolban]

Dashy,

Maybe the account that shows 0 quota doesn't have IAM permissions to use Vertex AI but the other account does?

[dashy]

Hi, I was able to enable all Vertex AIs with no issues. Can you help me with the steps on how I can check whether Vertex AI is enabled on the IAM level? 

[mchrestkha1]

thanks for sharing! looking into it.  should have a response tomorrow.

[mchrestkha1]

@dashy can you try again? I believe we increased to 30 queries per minute.

[dashy]

It's still at 0 for the enterprise account. For the individual account it's 30.

[adamsb]

It's the same for my enterprise account. All the quotas are 0 and I can't request more.

[wenshenjun]

I'm having the same issue; could you increase my limit?

[calvinw]

Hello all:

I see the same 0 quota with the inability to change it. Says contact sales. My other google account at work does have the quota set to 30. 

[valbeloru]

I have the same problem.  I was not able even to try it.
ERROR. Quota exceeded for aiplatform.googleapis.com/online_prediction_requests_per_base_model with base model: chat-bison
I even don't have Vertex Service on the Quotas page. 

Correction. Vertex has appeared. Now just the quota is zero. 

[JasonCW]

I'm having similar issue.  I created a free account to evaluate the process of finetuning a language model using Generative AI Studio.  I have a valid dataset JSONL file and I'm set to use us-central 1, but no matter what I do, every attempt always fails in the pipeline at the 'large-language-model-tuning' step with error AiPlatformException: code=RESOURCE_EXHAUSTED messsage=The following quota metrics exceed quota limits.  Oddly enough further down in the large complex error stack I notice in the path in the Execution name: that.../locations/europe-west4 is always mentioned even though all my settings are set to us-central1, even shows location as us-central1 in the Input Parameters.

I guess my ask is;  Is there some kind of error going on, or is using Generative AI Studio's Tuning feature just not something you are permitted evaluate in Free mode?

[sgardoll]

 I'm having exactly the same issue with a number of my Cloud projects. No matter what I do I always end up with the error:

"Error message: "Quota exceeded for aiplatform.googleapis.com/online_prediction_requests_per_base_model with base model: text-bison. Please submit a quota increase request.""

[animus_tom]

Same for us - in a project with billing attached where we're successfully using Vision AI and others

[PatrickBraunst]

I am having the same problem. I could not even try out the models. I just get: 

And my quotas look like:

[Izzy]

This same problem.

 

[oiwejdsd]

I contacted Sales support, but they told me they can't help since this is a newly launched feature and referred me to technical support.

I encourage everyone to raise this issue with technical support and also submit Feedback on Generative AI Studio such that we get heard.

[mchrestkha1]

thanks everyone for flagging.  We're currently looking into it and will report back.

[animus_tom]

Appreciate it @mchrestkha1 🙏

[Siewming8]

Me too. Is that mean free quota finished, 

[marshprice]

I'm having same issue/error. I have a paid account with billing account attached. And I looked at my quotas and I have "Online prediction requests per base model per minute per region per base_model" for "text-bison" set at 30 in us-cental1. So what do I look at next to fix the error that's preventing me from exploring Gen AI Studio? 

[joosh]

I submitted ~5 successful prompts, then started getting this error. Persisted through refresh, page changes, etc. However, it went away when I logged out and back in. Worth a try for those still having the issue.

Update: the issue came back after only a couple more attempts, and now does not go with a logout cycle.

[collasanta]

are you guys from europe? 

i saw that google havent allowed people from some countrys to use palm2, maybe this is due to a restriction based on your account location... Im from Brazil and its also not in the Allowed countrys, and i am getting the same error.

Up

[PatrickBraunst]

Yes, I am from Europe, but I mean, I am not trying to deploy PALM in Europe. I am fine with it being deployed in the US, as I am serving users from the US. If I couldn't test these products, simply because I personally am sitting in Europe, that would be a bad decision by Google in my opinion then.

[marshprice]

Today (or yesterday), Google Cloud released several Google Cloud Skills Boost labs (Qwiklabs) that used the same notebooks and console instructions  I was having issues with previously (so I don't think it was a quota issue). I got a similar error message when running notebook cells via GCSB, with a little phrase added ""Please try again later with backoff".  After consulting Bard on what that meant, Bard indicated that perhaps Vertex AI service is experiencing temporary high demand. I eventually was able to run all the code cells and/or console activities, but had to wait and resubmit some. I suggest checking out Google Cloud Skills Boost for labs/quests on "Generative AI", to explore using Vertex AI's Generative AI Studio, I found it helpful. 

[TL1999]

I'm having the same issue

[faroukegypt2]

I'm having the same issue

0 Likes

[mchrestkha1 · Accepted solution]

UPDATE: We have raised the default quotas for everyone.  This roll out may take a day to reach everyone so.  Thank you everyone for your patience and flagging this to us!

[Ch18b001]

aiplatform.googleapis.com

…internal.JobService.CreateCustomJob

…ects/601521391134/locations/europe-west4

service-601521391134@gcp-…

Hi @mchrestkha1 

Getting same error from long time tried all possibilities couldn't extend the quota limits. I'm using free tier credits.

com.google.cloud.ai.platform.common.errors.AiPlatformException: code=RESOURCE_EXHAUSTED, message=The following quota metrics exceed quota limits: aiplatform.googleapis.com/custom_model_training_cpus, cause=null

[ThomasVallee]

I am also still running into this issue when attempting to train a bison model.