Skip to main content

Google Vertex AI Configuration

Overview​

This guide explains how to enable Google Vertex AI models for use with AI/Run CodeMie. Google Vertex AI provides access to both native Google models (Gemini series) and partner models from providers like Anthropic (Claude).

Model Availability

Native Google Models: Available immediately after enabling Vertex AI API

Partner Models: Require explicit enablement in Model Garden

Prerequisites​

Before starting, ensure you have:

  • GCP Project: Active Google Cloud project with billing enabled
  • IAM Permissions: Vertex AI User or Vertex AI Administrator role
  • APIs Enabled: Vertex AI API must be enabled
  • Service Account: Service account with appropriate permissions (for programmatic access)

Step 1: Enable Vertex AI API​

1.1 Navigate to Vertex AI​

  1. Go to Google Cloud Console
  2. Select your project from the project dropdown
  3. In the navigation menu or search bar, find Vertex AI
  4. Click on Vertex AI to open the service

1.2 Enable the API​

If this is your first time using Vertex AI:

  1. You'll see a prompt to enable the Vertex AI API
  2. Click Enable or Enable API
  3. Wait for the API to be enabled (typically takes 1-2 minutes)

Step 2: Access Model Garden​

2.1 Navigate to Model Garden​

  1. In Vertex AI console, locate the left navigation menu
  2. Click Model Garden
  3. You'll see a catalog of available models organized by provider

2.2 Understanding Model Types​

Models in Vertex AI are categorized as:

Native Google Models (Available by Default):

  • Gemini 3 Pro
  • Gemini 2.5 Pro
  • Text Embeddings (text-embedding-005)

Partner Models (Require Enablement):

  • Anthropic Claude (Claude 4 series)

Step 3: Enable Partner Models​

3.1 Enable Claude Models (Example)​

To enable Anthropic Claude models on Vertex AI:

  1. In Model Garden, navigate to Partner models section or search for "Claude"
  2. Select the Claude model you want to enable (e.g., "Claude Opus 4.5")
  3. Review the model details page showing:
    • Model capabilities
    • Token limits
    • Supported data types
    • Pricing information
  4. Click Enable button at the top of the page
  5. Review and accept the terms of service
  6. Wait for enablement to complete