Supported Third-Party Provider Operations and Endpoints

Review a list of third-party REST providers and REST endpoints that are supported for various vector generation, summarization, text generation, and reranking operations.

Public or Remote REST Endpoint Providers

The supported publicly-hosted, third-party REST endpoint providers are:

Cohere
Google AI
Hugging Face
Oracle Cloud Infrastructure (OCI) Generative AI
OpenAI
Vertex AI

Local REST Endpoint Provider

You can use Ollama as a third-party REST endpoint provider, locally and privately on your Linux, Windows, and macOS systems.

Ollama is a free and open-source command-line interface tool that allows you to run open LLMs (such as Llama 3, Phi 3, Mistral, and Gemma 2) and embedding models (such as mxbai-embed-large, nomic-embed-text, or all-minilm). You can access Ollama using SQL and PL/SQL commands.

You can download and run the Ollama application from https://ollama.com/download. You can either install Ollama as a service that runs in the background or as a standalone binary with a manual install. For detailed installation-specific steps, see Quick Start in the Ollama Documentation.

REST Operations

These are the supported third-party REST operations and APIs along with their corresponding REST providers:

Operation	Provider	API
Generate embedding: Convert textual documents and images to one or more vector embeddings	For text input: All supported public and local providers For image input: Vertex AI	DBMS_VECTOR.UTL_TO_EMBEDDING and DBMS_VECTOR.UTL_TO_EMBEDDINGS DBMS_VECTOR_CHAIN.UTL_TO_EMBEDDING and DBMS_VECTOR_CHAIN.UTL_TO_EMBEDDINGS
Generate summary: Extract a brief and comprehensive summary from textual documents	All supported public and local providers	DBMS_VECTOR_CHAIN.UTL_TO_SUMMARY
Generate text: Retrieve a descriptive response for textual prompts and images through conversations with LLMs	For text input: All supported public and local providers For image input: Google AI, Hugging Face, OpenAI, Ollama, and Vertex AI	DBMS_VECTOR.UTL_TO_GENERATE_TEXT DBMS_VECTOR_CHAIN.UTL_TO_GENERATE_TEXT
Rerank results: Reassess and reorder search results to retrieve a more relevant output	Cohere and Vertex AI	DBMS_VECTOR.RERANK DBMS_VECTOR_CHAIN.RERANK

REST Endpoints

These are the supported REST endpoints for all third-party REST providers:

API	Provider	Endpoint
`UTL_TO_EMBEDDING` and `UTL_TO_EMBEDDINGS`	Cohere	`https://api.cohere.ai/v1/embed`
	Generative AI	`https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions/embedText`
	Google AI	`https://generativelanguage.googleapis.com/v1beta/models/`
	Hugging Face	`https://api-inference.huggingface.co/pipeline/feature-extraction/`
	Ollama	`http://localhost:11434/api/embeddings`
	OpenAI	`https://api.openai.com/v1/embeddings`
	Vertex AI	`https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT/locations/LOCATION/publishers/google/models/`
`UTL_TO_SUMMARY`	Cohere	`https://api.cohere.ai/v1/chat` `https://api.cohere.ai/v1/summarize`
	Generative AI	`https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions/chat`
	Google AI	`https://generativelanguage.googleapis.com/v1beta/models/`
	Hugging Face	`https://api-inference.huggingface.co/models/`
	Ollama	`http://localhost:11434/api/generate`
	OpenAI	`https://api.openai.com/v1/chat/completions` `https://api.openai.com/v1/completions`
	Vertex AI	`https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT/locations/LOCATION/publishers/google/models/`
`UTL_TO_GENERATE_TEXT`	Cohere	`https://api.cohere.ai/v1/chat` `https://api.cohere.ai/v1/generate`
	Generative AI	`https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions/chat`
	Google AI	`https://generativelanguage.googleapis.com/v1beta/models/`
	Hugging Face	`https://api-inference.huggingface.co/models/`
	Ollama	`http://localhost:11434/api/generate`
	OpenAI	`https://api.openai.com/v1/chat/completions` `https://api.openai.com/v1/completions`
	Vertex AI	`https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT/locations/LOCATION/publishers/google/models/`
`RERANK`	Cohere	`https://api.cohere.com/v1/rerank`
`RERANK`	Vertex AI	`https://discoveryengine.googleapis.com/v1/projects/PROJECT/locations/global/rankingConfigs/default_ranking_config:rank`

Models Supported for Generative AI

These are the third-party models that are supported to use with Generative AI corresponding to each REST API:

API Model

API	Model
`UTL_TO_EMBEDDING` and `UTL_TO_EMBEDDINGS`	cohere.embed-english-v3.0 cohere.embed-multilingual-v3.0 cohere.embed-english-light-v3.0 cohere.embed-multilingual-light-v3.0
`UTL_TO_SUMMARY`	cohere.command-r-16k cohere.command-r-plus meta.llama-3.1-70b-instruct meta.llama-3.1-405b-instruct
`UTL_TO_GENERATE_TEXT`	cohere.command-r-16k cohere.command-r-plus meta.llama-3.1-70b-instruct meta.llama-3.1-405b-instruct

UTL_TO_EMBEDDING

and

UTL_TO_EMBEDDINGS

cohere.embed-english-v3.0

cohere.embed-multilingual-v3.0

cohere.embed-english-light-v3.0

cohere.embed-multilingual-light-v3.0

UTL_TO_SUMMARY

cohere.command-r-16k

cohere.command-r-plus

meta.llama-3.1-70b-instruct

meta.llama-3.1-405b-instruct

UTL_TO_GENERATE_TEXT

cohere.command-r-16k

cohere.command-r-plus

meta.llama-3.1-70b-instruct

meta.llama-3.1-405b-instruct

Parent topic: Work with LLM-Powered APIs and Retrieval Augmented Generation