Data source - Elasticsearch (preview) (classic)

Applies only to: Foundry (classic) portal. This article isn't available for the new Foundry portal. Learn more about the new portal.

Note

Links in this article might open content in the new Microsoft Foundry documentation instead of the Foundry (classic) documentation you're viewing now.

Important

Azure OpenAI On Your Data retired on October 14, 2026, and is no longer supported by Microsoft.

For an Azure AI Search-backed application, follow Migrate Azure OpenAI On Your Data to Foundry IQ. This sample reuses your existing index and demonstrates application changes for retrieval, answers, and citations. Migration isn't automatic. See also the notebook source on GitHub.

The sample's auto retrieval and answer synthesis (answerSynthesis) use 2026-08-01-preview. Review its prerequisites and access-control requirements and the Azure AI Search preview terms. Preview features aren't recommended for production. For generally available alternatives, use minimal, extractive knowledge base retrieval with 2026-04-01 or classic RAG, with application-managed answer generation. These paths aren't feature-equivalent replacements.

Before retirement, Azure OpenAI On Your Data supported the following models:

GPT‑4o (versions 2024‑05‑13, 2024‑08‑06, and 2024‑11‑20)
GPT-4o-mini (version 2024-07-18)

For corpus Q&A, call a Foundry IQ knowledge base directly. Add Foundry Agent Service for additional tools, actions, or workflows. See Choose the migration path. For the agent-based path, see Connect agents to Foundry IQ knowledge bases.

The configurable options for Elasticsearch when using Azure OpenAI On Your Data. This data source is supported starting in API version 2024-02-15-preview.

Name	Type	Required	Description
`parameters`	Parameters	True	The parameters to use when configuring Elasticsearch.
`type`	string	True	Must be `elasticsearch`.

Parameters

Name	Type	Required	Description
`endpoint`	string	True	The absolute endpoint path for the Elasticsearch resource to use.
`index_name`	string	True	The name of the index to use in the referenced Elasticsearch.
`authentication`	One of KeyAndKeyIdAuthenticationOptions, EncodedApiKeyAuthenticationOptions	True	The authentication method to use when accessing the defined data source.
`embedding_dependency`	One of DeploymentNameVectorizationSource, EndpointVectorizationSource, ModelIdVectorizationSource	False	The embedding dependency for vector search. Required when `query_type` is `vector`.
`fields_mapping`	FieldsMappingOptions	False	Customized field mapping behavior to use when interacting with the search index.
`in_scope`	boolean	False	Whether queries should be restricted to use of indexed data. Default is `True`.
`query_type`	QueryType	False	The query type to use with Elasticsearch. Default is `simple`
`role_information`	string	False	Give the model instructions about how it should behave and any context it should reference when generating a response. You can describe the assistant's personality and tell it how to format responses.
`strictness`	integer	False	The configured strictness of the search relevance filtering. The higher of strictness, the higher of the precision but lower recall of the answer. Default is `3`.
`top_n_documents`	integer	False	The configured top number of documents to feature for the configured query. Default is `5`.

Authentication Options

Azure OpenAI On Your Data supports multiple authentication types:

Key and key ID authentication options

The authentication options for Azure OpenAI On Your Data when using an API key.

Name	Type	Required	Description
`key`	string	True	The Elasticsearch key to use for authentication.
`key_id`	string	True	The Elasticsearch key ID to use for authentication.
`type`	string	True	Must be `key_and_key_id`.

Encoded API key authentication options

The authentication options for Azure OpenAI On Your Data when using an Elasticsearch encoded API key.

Name	Type	Required	Description
`encoded_api_key`	string	True	The Elasticsearch encoded API key to use for authentication.
`type`	string	True	Must be `encoded_api_key`.

Deployment name vectorization source

The details of the vectorization source, used by Azure OpenAI On Your Data when applying vector search. This vectorization source is based on an internal embeddings model deployment name in the same Azure OpenAI resource. This vectorization source enables you to use vector search without Azure OpenAI api-key and without Azure OpenAI public network access.

Name	Type	Required	Description
`deployment_name`	string	True	The embedding model deployment name within the same Azure OpenAI resource.
`type`	string	True	Must be `deployment_name`.

Endpoint vectorization source

The details of the vectorization source, used by Azure OpenAI On Your Data when applying vector search. This vectorization source is based on the Azure OpenAI embedding API endpoint.

Name	Type	Required	Description
`endpoint`	string	True	Specifies the resource endpoint URL from which embeddings should be retrieved. It should be in the format of `https://{YOUR_RESOURCE_NAME}.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME/embeddings`. The api-version query parameter isn't allowed.
`authentication`	ApiKeyAuthenticationOptions	True	Specifies the authentication options to use when retrieving embeddings from the specified endpoint.
`type`	string	True	Must be `endpoint`.

API key authentication options

The authentication options for Azure OpenAI On Your Data when using an API key.

Name	Type	Required	Description
`key`	string	True	The API key to use for authentication.
`type`	string	True	Must be `api_key`.

Model ID vectorization source

The details of the vectorization source, used by Azure OpenAI On Your Data when applying vector search. This vectorization source is based on Elasticsearch model ID.

Name	Type	Required	Description
`model_id`	string	True	Specifies the model ID to use for vectorization. This model ID must be defined in Elasticsearch.
`type`	string	True	Must be `model_id`.

Fields mapping options

Optional settings to control how fields are processed when using a configured Elasticsearch resource.

Name	Type	Required	Description
`content_fields`	string[]	False	The names of index fields that should be treated as content.
`vector_fields`	string[]	False	The names of fields that represent vector data.
`content_fields_separator`	string	False	The separator pattern that content fields should use. Default is `\n`.
`filepath_field`	string	False	The name of the index field to use as a filepath.
`title_field`	string	False	The name of the index field to use as a title.
`url_field`	string	False	The name of the index field to use as a URL.

Query type

The type of Elasticsearch retrieval query that should be executed when using it with Azure OpenAI On Your Data.

Enum Value	Description
`simple`	Represents the default, simple query parser.
`vector`	Represents vector search over computed data.

Examples

Prerequisites:

Configure the role assignments from the user to the Azure OpenAI resource. Required role: Cognitive Services OpenAI User.
Install Az CLI and run az login.
Define the following environment variables: AzureOpenAIEndpoint, ChatCompletionsDeploymentName, SearchEndpoint, IndexName, Key, KeyId.

export AzureOpenAIEndpoint=https://example.openai.azure.com/
export ChatCompletionsDeploymentName=turbo
export SearchEndpoint='https://example.eastus.azurecontainer.io'
export IndexName=testindex
export Key='***'
export KeyId='***'

Python 1.x
REST

Install the latest pip packages openai, azure-identity.

import os
from openai import AzureOpenAI
from azure.identity import DefaultAzureCredential, get_bearer_token_provider

endpoint = os.environ.get("AzureOpenAIEndpoint")
deployment = os.environ.get("ChatCompletionsDeploymentName")
index_name = os.environ.get("IndexName")
search_endpoint = os.environ.get("SearchEndpoint")
key = os.environ.get("Key")
key_id = os.environ.get("KeyId")

token_provider = get_bearer_token_provider(
    DefaultAzureCredential(), "https://ai.azure.com/.default")

client = AzureOpenAI(
    azure_endpoint=endpoint,
    azure_ad_token_provider=token_provider,
    api_version="2024-02-15-preview",
)

completion = client.chat.completions.create(
    model=deployment,
    messages=[
        {
            "role": "user",
            "content": "Who is DRI?",
        },
    ],
    extra_body={
        "data_sources": [
            {
                "type": "elasticsearch",
                "parameters": {
                    "endpoint": search_endpoint,
                    "index_name": index_name,
                    "authentication": {
                        "type": "key_and_key_id",
                        "key": key,
                        "key_id": key_id
                    }
                }
            }
        ]
    }
)

print(completion.model_dump_json(indent=2))


az rest --method POST \
 --uri $AzureOpenAIEndpoint/openai/deployments/$ChatCompletionsDeploymentName/chat/completions?api-version=2024-02-15-preview \
 --resource https://cognitiveservices.azure.com/ \
 --body \
'
{
    "data_sources": [
      {
        "type": "elasticsearch",
        "parameters": {
          "endpoint": "'$SearchEndpoint'",
          "index_name": "'$IndexName'",
          "authentication": {
            "type": "key_and_key_id",
            "key": "'$Key'",
            "key_id": "'$KeyId'"
          }
        }
      }
    ],
    "messages": [
      {
        "role": "user",
        "content": "Who is DRI?"
      }
    ]
}
'

Feedback

Was this page helpful?

Last updated on 2026-06-05