Supported AI models

To build generative AI applications with AlloyDB for PostgreSQL, you can register remote model endpoints using Model endpoint management. This lets you use SQL to interact with Large Language Models (LLMs) for tasks such as generating text embeddings, multimodal content, and semantic ranking.

This page provides a list of AI models that have built-in support or are pre-registered for use with AlloyDB. These AI models simplify your integration with Gemini Enterprise Agent Platform, OpenAI, and Anthropic. For more information about different model types, see Model type.

Models with built-in support

AlloyDB provides built-in support for several models from AI providers. For these models, AlloyDB handles the underlying integration details automatically, such as setting up default transform functions. This occurs when you register the model endpoint using its qualified name.

This section lists the qualified model names and types for the Agent Platform, OpenAI, and Anthropic models.

Agent Platform

Note: Agent Platform model versioning and lifecycle guidelines govern Agent Platform model support. For more information about versions of stable models, see Model versions and lifecycle.

Model type	Model name
text-embedding	text-embedding-004
text-embedding	text-embedding-005
text-embedding	text-embedding-preview-0815
text-embedding	text-multilingual-embedding-002
text-embedding	text-embedding-large-exp-03-07 ^*
multimodal_embedding	multimodalembedding@001
reranking	semantic-ranker-default-003

• The text-embedding-large-exp-03-07 model is available only in the us-central1 region.

OpenAI

Model type	Qualified model name
text-embedding	text-embedding-ada-002
text-embedding	text-embedding-3-small
text-embedding	text-embedding-3-large

Anthropic

Model type	Qualified model name
generic	claude-3-opus-20240229
generic	claude-3-sonnet-20240229
generic	claude-3-haiku-20240307

Pre-registered Agent Platform models

For the following Agent Platform models, AlloyDB provides pre-registration, which lets you call these models directly using their qualified names without requiring a manual registration step. This offers a streamlined way to use Agent Platform capabilities in your SQL queries.

The following table lists the model types and IDs of pre-registered Agent Platform models.

Note: Agent Platform model versioning and lifecycle guidelines govern Agent Platform model support. For more information, see Model versions and lifecycle.

Model type	Model ID
text-embedding	text-embedding-005
text-embedding	gemini-embedding-001
text-embedding	text-multilingual-embedding-002
multimodal-embedding	multimodalembedding
multimodal-embedding	multimodalembedding@001
reranking	semantic-ranker-512
reranking	semantic-ranker-default-002
reranking	semantic-ranker-default-003
reranking	semantic-ranker-default-004
reranking	semantic-ranker-fast-004
llm	gemini-2.0-flash
llm	gemini-2.5-flash
llm	gemini-2.0-flash-lite
llm	gemini-2.5-flash-lite
llm	gemini-2.5-pro