A retriever is an interface that returns documents given an unstructured query. It is more general than a vector store. A retriever does not need to be able to store documents, only to return (or retrieve) them. Retrievers can be created from vector stores, but are also broad enough to include other sources. Retrievers accept a string query as input and return a list ofDocumentation Index
Fetch the complete documentation index at: https://docs.langchain.com/llms.txt
Use this file to discover all available pages before exploring further.
Document objects as output.
Note that all vector stores can be cast to retrievers. Refer to the vector store integration docs for available vector stores.
This page lists custom retrievers, implemented via subclassing BaseRetriever.
Bring-your-own documents
The below retrievers allow you to index and search a custom corpus of documents.| Retriever | Self-host | Cloud offering | Package |
|---|---|---|---|
AmazonKnowledgeBasesRetriever | ❌ | ✅ | langchain-aws |
ElasticsearchRetriever | ✅ | ✅ | langchain-elasticsearch |
NVIDIARAGRetriever | ✅ | ❌ | langchain-nvidia-ai-endpoints |
VertexAISearchRetriever | ❌ | ✅ | langchain-google-community |
External index
The below retrievers will search over an external index (e.g., constructed from Internet data or similar).| Retriever | Source | Package |
|---|---|---|
ParallelSearchRetriever | Internet search via the Parallel Search API | langchain-parallel |
PerplexitySearchRetriever | Internet search via the Perplexity Search API | langchain-perplexity |
YouRetriever | Internet search | langchain-youdotcom |
All retrievers
Bedrock (Knowledge Bases)
Box
Cognee
Cohere reranker
Cohere RAG
Contextual AI Reranker
Dappier
Elasticsearch
Egnyte
Galaxia
Google Drive
Google Vertex AI Search
Graph RAG
GreenNode
IBM watsonx.ai
IMAP
Kinetica Vectorstore
LinkupSearchRetriever
Nebius
Nimble Extract
Nimble Search
NVIDIA RAG Blueprint
Parallel Search
Permit
Perigon
Perplexity Search
Pinecone Rerank
RAGatouille
SpiceDB
ValyuContext
Vectorize
You.com
Zotero
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.