AzureOpenAIEmbeddings: Reference Guide

What is AzureOpenAIEmbeddings?

AzureOpenAIEmbeddings is the LangChain wrapper for OpenAI embedding models served through Microsoft Azure OpenAI rather than OpenAI directly. The vectors it returns are identical in quality to OpenAIEmbeddings — it is the same text-embedding-3-small, text-embedding-3-large, and text-embedding-ada-002 models underneath — but requests are routed to your Azure resource, billed through your Azure subscription, and kept inside whatever region and network boundary your compliance team requires.

The key difference from the standard class is configuration. Azure addresses a named deployment, not a model: you create a deployment in the Azure portal, then pass its name through azure_deployment (or the legacy deployment_id), set azure_endpoint to your resource URL, and pin api_version to a dated value such as 2024-02-01. Authentication uses AZURE_OPENAI_API_KEY and AZURE_OPENAI_ENDPOINT environment variables, or an Azure AD token provider for keyless auth. Forgetting any one of endpoint, deployment, or api_version is the most common setup failure.

At the API surface it exposes the same embed_query and embed_documents methods as every other LangChain embeddings class, so it drops into a Chroma, Pinecone, or Qdrant retriever without code changes. embed_documents batches internally; for large corpora tune chunk_size to stay under your deployment token-per-minute quota, and set dimensions on the v3 models if you want shorter vectors to save storage. Crucially, the embedding model and deployment you index with must match the one you query with — mixing models silently corrupts similarity scores.

When to Use

You're on Azure or need Azure's enterprise features. Use AzureOpenAIEmbeddings for Azure deployments.

Use Cases

• • Azure deployments
• • Enterprise compliance
• • Private networks
• • Multi-region
• • Azure managed services
• • Regulatory compliance

Key Features

• ✓ Azure-hosted
• ✓ Private access
• ✓ Enterprise features
• ✓ Same models
• ✓ Regional deployment
• ✓ Managed scaling

When NOT to Use

Notes

Import

from langchain_openai import AzureOpenAIEmbeddings

Configuration

Usage Examples

from langchain_openai import AzureOpenAIEmbeddings

embeddings = AzureOpenAIEmbeddings(
    azure_deployment='text-embedding-3-small',
    azure_endpoint='https://my-resource.openai.azure.com/',
    api_version='2024-02-01',
)
vector = embeddings.embed_query('What is retrieval augmented generation?')
print(len(vector))

from langchain_openai import AzureOpenAIEmbeddings
from langchain_chroma import Chroma

embeddings = AzureOpenAIEmbeddings(
    azure_deployment='text-embedding-3-large',
    api_version='2024-02-01',
    chunk_size=256,
)
store = Chroma.from_documents(docs, embeddings)
hits = store.similarity_search('billing policy', k=4)

from azure.identity import DefaultAzureCredential, get_bearer_token_provider
from langchain_openai import AzureOpenAIEmbeddings

token_provider = get_bearer_token_provider(
    DefaultAzureCredential(),
    'https://cognitiveservices.azure.com/.default',
)
embeddings = AzureOpenAIEmbeddings(
    azure_deployment='text-embedding-3-small',
    azure_endpoint='https://my-resource.openai.azure.com/',
    api_version='2024-02-01',
    azure_ad_token_provider=token_provider,
)

Common Pitfalls

Alternative Embedding Models

Related LangChain References

Browse the full LangChain API reference index to explore more classes, methods, and decorators, or start with the LangChain introduction tutorial for end-to-end context on building with AzureOpenAIEmbeddings and the wider framework.

AzureOpenAIEmbeddings FAQ