Register and call remote AI models in Cloud SQL overview

This page describes how to register an AI model endpoint and invoking predictions with model endpoint management in Cloud SQL. To use AI models in production environments, see Build generative AI applications using Cloud SQL and Work with vector embeddings.

Overview

Model endpoint management lets you register a model endpoint, manage model endpoint metadata in your Cloud SQL instance, and then interact with the models using SQL queries. Cloud SQL provides the google_ml_integration extension that includes functions to add and register the model endpoint metadata related to the models. You can use these models to generate vector embeddings or invoke predictions.

You can register the following model types by using model endpoint management:

• Vertex AI text embedding models.
• Custom-hosted text embedding models hosted in networks within Google Cloud.
• Generic models with a JSON-based API. Examples of these models include the following:
  • The gemini-pro model from the Vertex AI Model Garden
  • The open_ai model for OpenAI models
  • Models hosted in networks within Google Cloud

How it works

You can use model endpoint management to register a model endpoint that complies to the following:

• The model input and output support the JSON format.
• You can use the REST protocol to call the model.

When you register a model endpoint with model endpoint management, model endpoint management registers each endpoint with a unique model ID as a reference to the model. You can use this model ID to query models, as follows:

• Generate embeddings to translate text prompts to numerical vectors. You can store generated embeddings as vector data when the pgvector extension is enabled in the database. For more information, see Query and index embeddings with pgvector.

• Invoke predictions to call a model using SQL within a transaction.

Your applications can manage their model endpoints using the google_ml_integration extension. This extension provides the following SQL functions:

• google_ml.create_model(): registers the model endpoint that's used in the prediction or embedding function
• google_ml.create_sm_secret(): uses secrets in Google Cloud Secret Manager, where the API keys are stored
• google_ml.embedding(): generates text embeddings
• google_ml.predict_row(): generates predictions when you call generic models that support the JSON input and output formats

Key concepts

Before you start using model endpoint management, understand the concepts required to connect to and use the models.

Model provider

Model provider is the supported model hosting providers. The following table shows the model provider value you must set based on the model provider you use:

The default model provider is custom.

Model types

Model types are the types of the AI model. When you register a model endpoint, you can set the text-embedding or generic model types for the endpoint.

Text embedding models with built-in support

Model endpoint management provides built-in support for all versions of the textembedding-gecko model. To register these model endpoints, use the google_ml.create_model() function. Cloud SQL sets up default transform functions for these models automatically.

The model type for these models is text-embedding.

Other text embedding models

For other text embedding models, you need to create transform functions to handle the input and output formats that the model supports. Optionally, you can use the HTTP header generation function that generates custom headers required by your model.

The model type for these models is text-embedding.

Generic models

Model endpoint management also supports registering of all other model types apart from text embedding models. To invoke predictions for generic models, use the google_ml.predict_row() function. You can set model endpoint metadata, such as a request endpoint and HTTP headers that are specific to your model.

You cannot pass transform functions when you register a generic model endpoint. Ensure that when you invoke predictions the input to the function is in the JSON format, and that you parse the JSON output to derive the final output.

The model type for these models is generic. Because generic is the default model type, if you register model endpoints for this type, then setting the model type is optional.

Authentication methods

You can use the google_ml_integration extension to specify different authentication methods to access your model. Setting these methods is optional and is required only if you need to authenticate to access your model.

For Vertex AI models, the Cloud SQL service account is used for authentication. For other models, the API key or bearer token that is stored as a secret in the Secret Manager can be used with the google_ml.create_sm_secret() SQL function.

The following table shows the authentication methods that you can set:

Prediction functions

The google_ml_integration extension includes the following prediction functions:

google_ml.embedding()

Calls a registered text embedding model endpoint to generate embeddings. It includes built-in support for the textembedding-gecko model by Vertex AI.

For text embedding models without built-in support, the input and output parameters are unique to a model and need to be transformed for the function to call the model. Create a transform input function to transform input of the prediction function to the model specific input, and a transform output function to transform model specific output to the prediction function output.

google_ml.predict_row()

Calls a registered generic model endpoint, if the endpoint supports JSON-based APIs to invoke predictions.

Transform functions

Transform functions modify the input to a format that the model understands, and convert the model response to the format that the prediction function expects. The transform functions are used when registering the text-embedding model endpoint without built-in support. The signature of the transform functions depends on the prediction function for the model type.

You cannot use transform functions when registering a generic model endpoint.

The following shows the signatures for the prediction function for text embedding models:

// define custom model specific input/output transform functions.
CREATE OR REPLACE FUNCTION input_transform_function(model_id VARCHAR(100), input_text TEXT) RETURNS JSON;

CREATE OR REPLACE FUNCTION output_transform_function(model_id VARCHAR(100), response_json JSON) RETURNS real[];

For more information about how to create transform functions, see Transform functions example.

HTTP header generation function

The HTTP header generation function generates the output in JSON key value pairs that are used as HTTP headers. The signature of the prediction function defines the signatures of the header generation function.

The following example shows the signature for the google_ml.embedding() prediction function:

CREATE OR REPLACE FUNCTION generate_headers(model_id VARCHAR(100), input TEXT) RETURNS JSON;

For the google_ml.predict_row() prediction function, the signature is as follows:

CREATE OR REPLACE FUNCTION generate_headers(model_id VARCHAR(100), input JSON) RETURNS JSON;

For more information about how to create a header generation function, see Header generation function example.

Limitations

• To use AI models with your Cloud SQL instance, the maintenance version of your instance must be R20240910.01_02 or later. To upgrade your instance to this version, see Self-service maintenance.

What's next

• Set up authentication for model providers.
• Register a model endpoint with model endpoint management.
• Learn about the model endpoint management reference.