Google Search is not working

[MOzcelik14]

Whenever I try to access real-time information, I get this error message:

System.Net.Http.HttpRequestException: 'Request failed
Request failed with Status Code: TooManyRequests

{
  "error": {
    "code": 429,
    "message": "Resource has been exhausted (e.g. check quota).",
    "status": "RESOURCE_EXHAUSTED"
  }
}

var model = googleAI.GenerativeModel(model: Model.Gemini20Pro, systemInstruction: systemInstruction);
model.UseGoogleSearch = true;
var response = await model.GenerateContent(prompt);

And this is my code.

[jochenkirstaetter]

Hi @MOzcelik14

Thanks for reaching out. The HTTP status code 429 as well as 503 is returned by the Gemini API when they have capacity issues. That's nothing to be done on the client side and your code is correct.

See test cases

• Generate_Content_with_Google_Search
• Generate_Content_with_GoogleSearch_ResponseModalities

for comparison.

Happens to me and others regularly, as you can see in the Build with Google AI Forum.

Either wait a bit or use another model and try it again. Switching models usually helps on my side.

Cheers.

[jochenkirstaetter]

Hi @MOzcelik14

Are you still observing this issue?
Or can I consider to close it?

Cheers.

[MOzcelik14]

hi @jochenkirstaetter

doesn't the 1.5 model have a grounding feature?

[MOzcelik14]

I am currently trying to develop an artificial intelligence assistant for the faculty I study at. Our dean says he wants to access real-time information. When I use the 2.0 Pro model, I get the 429 error message after 1 or 2 prompts. And I think there is no grounding feature in any model of the 1.5.

[jochenkirstaetter]

Hello @MOzcelik14

Yes, the Gemini 1.5 models also support Google Search and Grounding (using Vertex AI).

Kindly have a look at the tests. There are a few examples. Have you tried the Gemini 2.0 Flash models already?

Cheers, JoKi

[jochenkirstaetter]

Hi @MOzcelik14

You have a choice:

• Latest: gemini-2.0-flash
  

• Stable: gemini-2.0-flash-001
  

Check whether the stable model behaves less problematic regarding the quota.
Reference: https://ai.google.dev/gemini-api/docs/models/gemini#gemini-2.0-flash

Or using Vertex AI, as previously suggested. There you have access to Gemini 1.5 models, too.
Reference: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/grounding

This SDK provides you with both options. Hope this helps.

[MOzcelik14]

How do I use the Vertex AI with Mscc.GenerativeAI then? I didn't understand how to set it up. Sorry for bothering you continously by the way.

[jochenkirstaetter]

Hi @MOzcelik14

I read on the Discord server of the Google Developer Community that there might have been update by the Gemini team recently.

Maybe this already improved your scenario.

Vertex AI

You can use this package the same way. It's the first few lines to initialize the model which are slightly different because of the differences between Google AI and Vertex AI.

In case that you've never used Google Cloud before, you can check out the so-called "express mode" which is still based on using an API key, however different to your existing one. Start here: https://cloud.google.com/vertex-ai/generative-ai/docs/start/express-mode/overview - kindly check the Available models and rate limits in express mode.

If that's not enough then you could either graduate from express mode or directly sign up for an account on Google Cloud. Then create a project and use this SDK with your projectId and a location (or region).

using Mscc.GenerativeAI;
// Vertex AI with OAuth. Use `gcloud auth application-default print-access-token` to get the access token.
var vertex = new VertexAI(projectId: projectId, region: region);
var model = vertex.GenerativeModel(model: Model.Gemini15Pro);
model.AccessToken = accessToken;

If you are going to deploy your solution to Google Cloud I suggest using the package Mscc.GenerativeAI.Google to handle implicit authentication using the Google Client Library.

More information also here: https://cloud.google.com/docs/authentication/provide-credentials-adc#local-dev