How to beginner · 3 min read

How to create embeddings for documents in LangChain

Q: How to create embeddings for documents in LangChain

Use OpenAIEmbeddings from langchain_openai to create embeddings for documents in LangChain. Load your documents with a loader like TextLoader, then call embedder.embed_documents() on the document texts to get vector embeddings.

Quick answer

Use OpenAIEmbeddings from langchain_openai to create embeddings for documents in LangChain. Load your documents with a loader like TextLoader, then call embedder.embed_documents() on the document texts to get vector embeddings.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install langchain_openai langchain_community

Setup

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

Run pip install langchain_openai langchain_community
Set environment variable OPENAI_API_KEY with your OpenAI API key

bash

pip install langchain_openai langchain_community

Step by step

This example loads a text document, creates embeddings using OpenAI's text-embedding-3-large model via LangChain, and prints the resulting vectors.

python

import os
from langchain_openai import OpenAIEmbeddings
from langchain_community.document_loaders import TextLoader

# Load document text
loader = TextLoader("example.txt")
docs = loader.load()

# Initialize OpenAI embeddings client
embedder = OpenAIEmbeddings(api_key=os.environ["OPENAI_API_KEY"], model_name="text-embedding-3-large")

# Extract texts from documents
texts = [doc.page_content for doc in docs]

# Generate embeddings
embeddings = embedder.embed_documents(texts)

print(f"Generated {len(embeddings)} embeddings.")
print("First embedding vector sample:", embeddings[0][:5])

output

Generated 1 embeddings.
First embedding vector sample: [0.0023, -0.0017, 0.0045, 0.0031, -0.0028]

Common variations

Use other loaders like PyPDFLoader for PDFs.
Switch to embedder.embed_query() for single text queries.
Use different embedding models by changing model_name.
Async embedding calls are supported with LangChain async methods.

Troubleshooting

If you get authentication errors, verify your OPENAI_API_KEY environment variable is set correctly.
For rate limit errors, consider retrying after some time or reducing batch size.
If embeddings are empty, check that documents are loaded properly and contain text.

✅

Key Takeaways

Use OpenAIEmbeddings from langchain_openai to generate document embeddings easily.
Load documents with loaders like TextLoader or PyPDFLoader before embedding.
Pass a list of document texts to embed_documents() for batch embedding.
Set your OpenAI API key in os.environ["OPENAI_API_KEY"] to authenticate.
You can switch embedding models by changing the model_name parameter.

Verified 2026-04 · text-embedding-3-large

Verify ↗