The idea is:
Add support for an Azure-hosted Cohere reranker sub-node in @n8n/nodes-langchain, including dedicated credentials for Azure Cohere rerank API access and wiring it into vector-store reranking flows.
My use case:
I’m building RAG workflows where retrieved chunks need reranking before final answer generation. My infrastructure uses Azure-hosted Cohere endpoints (not direct Cohere public API), so I need a native n8n node/credential setup that works with Azure base URLs and existing retriever/vector-store patterns.
I think it would be beneficial to add this because:
n8n currently supports Cohere reranking, but Azure-hosted Cohere is a common enterprise deployment path with different endpoint configuration requirements. Native support reduces custom-code workarounds, makes RAG quality improvements easier to adopt, and aligns with n8n’s existing AI sub-node UX for retrievers/vector stores.Any resources to support this?
- n8n Reranker Cohere docs:
Reranker Cohere | n8n Docs - n8n Vector Store Retriever docs:
Vector Store Retriever node documentation | n8n Docs - n8n Simple Vector Store docs (rerank option and connection pattern):
Simple Vector Store node documentation | n8n Docs - Cohere model/docs reference:
An Overview of Cohere's Models | Cohere
Are you willing to work on this?
Yes. I’ve already implemented this in my fork and can open/maintain the PR.