ServiceTier when using Vertex AI

[symphony-enrico]

When using Vertex AI, if I try to set the ServiceTier (e.g: FLEX), I receive the following error:

{
  "error": {
    "code": 400,
    "message": "Invalid value at 'service_tier' (type.googleapis.com/google.cloud.aiplatform.v1beta1.ServiceTier), \"flex\"",
    "status": "INVALID_ARGUMENT",
    "details": [
      {
        "@type": "type.googleapis.com/google.rpc.BadRequest",
        "fieldViolations": [
          {
            "field": "service_tier",
            "description": "Invalid value at 'service_tier' (type.googleapis.com/google.cloud.aiplatform.v1beta1.ServiceTier), \"flex\""
          }
        ]
      }
    ]
  }
}

This problem seems very similar to: googleapis/js-genai#1468 in particular see googleapis/js-genai#1468 (comment)

[hemasekhar-p]

Hello @symphony-enrico, thank you for reporting this issue. To assist us in effectively investigating and reproducing it, please provide a minimal, runnable code example, including the complete error logs and sdk version. these things would be very helpful. Thanks!

[symphony-enrico]

Thanks @hemasekhar-p
I tried with latest version 1.52.0 now.

Code example:

Login with:

gcloud auth login --update-adc

Replace PUT-YOUR-PROJECT and run:

package org.example;

import com.google.genai.Client;
import com.google.genai.types.*;

public class GeminiApp {
    public static void main(String[] args) {
        Client client = Client.builder()
                .project("PUT-YOUR-PROJECT")
                .location("global")
                .vertexAI(true)
                .build();

        GenerateContentConfig contentConfig = GenerateContentConfig.builder()
                .temperature(1f)
                .serviceTier(ServiceTier.Known.PRIORITY)
                .build();

        GenerateContentResponse response = client.models
                .generateContent("gemini-3.1-pro-preview", "Who are you?", contentConfig);

        String responseText = response.text();
        System.out.println(responseText);
    }
}

Observed result:

Exception in thread "main" com.google.genai.errors.ClientException: 400 . Invalid value at 'service_tier' (type.googleapis.com/google.cloud.aiplatform.v1beta1.ServiceTier), "priority"
	at com.google.genai.errors.ApiException.throwFromResponse(ApiException.java:94)
	at com.google.genai.HttpApiResponse.getBody(HttpApiResponse.java:37)
	at com.google.genai.Models.processResponseForPrivateGenerateContent(Models.java:5592)
	at com.google.genai.Models.privateGenerateContent(Models.java:5658)
	at com.google.genai.Models.generateContent(Models.java:7194)
	at com.google.genai.Models.generateContent(Models.java:7268)
	at org.example.GeminiApp.main(GeminiApp.java:20)

The error is caused by adding the configuration

  .serviceTier(ServiceTier.Known.PRIORITY)

Because service tier configuration is added into payload, that is not the correct syntax for VertexAI, but instead into headers.
Just for reference, if I remove serviceTier but I put manually the required headers, it works:

package org.example;

import com.google.genai.Client;
import com.google.genai.types.GenerateContentConfig;
import com.google.genai.types.GenerateContentResponse;
import com.google.genai.types.HttpOptions;

import java.util.Map;

public class GeminiApp2 {
    public static void main(String[] args) {
        Client client = Client.builder()
                .project("sym-dev-use4-data-01")
                .location("global")
                .vertexAI(true)
                .httpOptions(HttpOptions.builder().headers(Map.of(
                        "X-Vertex-AI-LLM-Request-Type", "shared",
                        "X-Vertex-AI-LLM-Shared-Request-Type", "priority")).build())
                .build();

        GenerateContentConfig contentConfig = GenerateContentConfig.builder()
                .temperature(1f)
                .build();

        GenerateContentResponse response = client.models
                .generateContent("gemini-3.1-pro-preview", "Who are you?", contentConfig);

        String responseText = response.text();
        System.out.println(responseText);
        System.out.println("ServiceTier: " + response.usageMetadata().get().trafficType().get());
    }
}

And the result are the one expected (and the serviceTier is taken into account too):

I am Gemini, a large language model, trained by Google. How can I help you today?
ServiceTier: ON_DEMAND_PRIORITY

[eljefe6a]

I implemented this without Vertex AI, and I didn't get an error.

[symphony-enrico]

I implemented this without Vertex AI, and I didn't get an error.

The error happens ONLY with Vertex AI, because the way of passing ServiceTier parameter is different (Vertex Ai expected to have specific headers, NOT data in payload).

Same bug for js-genai, and it is better explained than my explanation: googleapis/js-genai#1468 (comment)

[hemasekhar-p]

@symphony-enrico, Thank you for providing the required details and complete reproducible code. I am able to reproduce this behavior and will file a bug internally to track it.

[jaycee-li]

@symphony-enrico Thanks for the feedback. The Vertex backend will soon support ServiceTier in the payload to enable the following CUJ:

        GenerateContentConfig contentConfig = GenerateContentConfig.builder()
                .serviceTier(ServiceTier.Known.PRIORITY)
                .build();

For now, please proceed with the custom header workaround as you noted above. Thanks for the patience!

[kalyanburam]

hi @symphony-enrico the workaround with custom header is not working for gemini-2.5-flash-lite or gemini-2.5-flash models. Is it a known bug or behavior?