How to beginner · 3 min read

How to use OpenAI embeddings in LangChain

Q: How to use OpenAI embeddings in LangChain

Use the OpenAIEmbeddings class from langchain_openai to generate embeddings with OpenAI models like o1-mini. Instantiate it with your API key from os.environ, then embed text by calling embed_documents or embed_query. This integrates seamlessly with LangChain vectorstores for semantic search.

Quick answer

Use the OpenAIEmbeddings class from langchain_openai to generate embeddings with OpenAI models like o1-mini. Instantiate it with your API key from os.environ, then embed text by calling embed_documents or embed_query. This integrates seamlessly with LangChain vectorstores for semantic search.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install langchain_openai openai>=1.0

Setup

Install the required packages and set your OpenAI API key as an environment variable.

Install LangChain OpenAI bindings and OpenAI SDK: pip install langchain_openai openai
Set your API key in your shell: export OPENAI_API_KEY='your_api_key_here' (Linux/macOS) or setx OPENAI_API_KEY "your_api_key_here" (Windows)

bash

pip install langchain_openai openai

Step by step

This example shows how to create an OpenAIEmbeddings instance, embed a list of documents, and embed a query string.

python

import os
from langchain_openai import OpenAIEmbeddings

# Initialize embeddings client with API key from environment
embeddings = OpenAIEmbeddings(openai_api_key=os.environ["OPENAI_API_KEY"], model_name="o1-mini")

# Example documents to embed
texts = [
    "LangChain makes working with LLMs easier.",
    "OpenAI provides powerful embedding models.",
    "Embeddings convert text into vectors for similarity search."
]

# Generate embeddings for documents
vector_representations = embeddings.embed_documents(texts)
print("Document embeddings shape:", len(vector_representations), "vectors")

# Generate embedding for a query
query = "How to use embeddings with LangChain?"
query_vector = embeddings.embed_query(query)
print("Query embedding vector length:", len(query_vector))

output

Document embeddings shape: 3 vectors
Query embedding vector length: 1536

Common variations

You can use different OpenAI embedding models by changing the model_name parameter, such as o1 or o1-mini. For asynchronous usage, LangChain currently does not support async embeddings directly, so run embedding calls in a thread or process pool. You can also combine OpenAIEmbeddings with vectorstores like FAISS or Chroma for semantic search.

Model name	Description
o1-mini	Smaller, faster embedding model with good quality
o1	Larger embedding model with higher accuracy

Troubleshooting

If you get authentication errors, verify your OPENAI_API_KEY environment variable is set correctly. If embeddings return empty or all zeros, check your network connection and API usage limits. For unexpected vector sizes, confirm you are using a supported model name. Use print statements to debug returned embeddings shapes.

✅

Key Takeaways

Use OpenAIEmbeddings from langchain_openai with your API key from os.environ.
Call embed_documents for batch text and embed_query for single queries to get vector embeddings.
Change model_name to switch between embedding models like o1-mini or o1.
Combine embeddings with LangChain vectorstores for semantic search and retrieval.
Check environment variables and model names carefully to avoid common errors.

Verified 2026-04 · o1-mini, o1

Verify ↗