Deployments and endpoints

Google and Partner models and generative AI features on Vertex AI are exposed as specific regional endpoints and a global endpoint. Global endpoints cover the entire world and provide higher availability and reliability than single regions.

Note that model endpoints don't guarantee region availability or in-region ML processing. For information about data residency, see Data residency .

Global endpoint

Selecting a global endpoint for your requests can improve overall availability while reducing resource exhausted (429) errors. Don't use the global endpoint if you have ML processing requirements, because you can't control or know which region your ML processing requests are sent to when a request is made.

Supported models

Usage of the global endpoint is supported for the following Google models in specified regions. For details about which regions support the global endpoint, see the Globaltab in the Google model endpoint locations table .

• Gemini 2.5 Flash Image Preview
• Gemini 2.5 Flash-Lite
• Gemini 2.5 Pro
• Gemini 2.5 Flash
• Gemini 2.0 Flash
• Gemini 2.0 Flash-Lite

For information about global endpoint availability for partner models, see the Globaltab in the Google Cloud partner model endpoint locations table .

Use the global endpoint

To use the global endpoint, exclude the location from the endpoint name and configure the location of the resource to global . For example, the following is global endpoint URL:

 https:// aiplatform.googleapis.com/v1/projects/test-project/locations/ global/publishers/google/models/gemini-2.0-flash-001:generateContent 

For the Google Gen AI SDK , create a client that uses the global location:

  client 
 = 
 genai 
 . 
 Client 
 ( 
 vertexai 
 = 
 True 
 , 
 project 
 = 
 ' PROJECT_ID 
' 
 , 
 location 
 = 
 'global' 
 ) 
 

For the Vertex AI SDK for Python , initialize the SDK using the global location:

  import 
  
  vertexai 
 
 from 
  
 vertexai.generative_models 
  
 import 
  GenerativeModel 
 
  vertexai 
 
 . 
 init 
 ( 
 project 
 = 
 ' PROJECT_ID 
' 
 , 
 location 
 = 
 'global' 
 ) 
 

Limitations

The following capabilities are not available when using the global endpoint:

• Tuning
• Batch prediction
• Retrieval-augmented generation (RAG) corpus (RAG requests are supported)

Usage of the global endpoint with Provisioned Throughput is available only for the following models:

Google model endpoint locations

Google model endpoints for Generative AI on Vertex AI are available in the following regions.

* Region is available only while using Single Zone Provisioned Throughput

+ Supervised fine tuning isn't supported in this region.

Google Cloud partner model endpoint locations

Google serves requests from the region that you specified. For some models, Google also offers a global endpoint to improve overall availability and reduce error rates. The global endpoint can have a separate set of quotas from the regional endpoint and doesn't support data residency requirements. For more information, see the "Regional and global endpoint" section in Vertex AI partner models for MaaS .

Partner model endpoints for Generative AI on Vertex AI are available in the following regions:

What's next

• For a notebook tutorial that demonstrates the global endpoint, see Intro to the Vertex AI global endpoint .
• Learn more about Generative AI on Vertex AI data residency .
• Learn about Google Cloud regions .
• Learn more about security controls by feature .
• Learn about the models that provide Generative AI on Vertex AI support. See Generative AI foundational model reference .
• Learn about Vertex AI locations .