Description

n8n-nodes-reranker-vllm

Use vLLM Reranker to reorder retrieved documents by their relevance to a given query.
This node leverages locally served reranker models (via vLLM to perform cross-encoder–style relevance scoring.

Built on the official n8n Community Node Starter, so you can develop, lint, and ship confidently.

✨ Supported Models

This node currently supports Qwen3 Rerankers by @Qwen:

• Qwen/Qwen3-Reranker-8B
• Qwen/Qwen3-Reranker-4B
• Qwen/Qwen3-Reranker-0.6B

All variants and quantizations (e.g., :Q5_K_M, :Q8_0) are automatically detected.

If you’d like to request support for other reranker models, please open a GitHub issue or feature request so it can be added to the whitelist.

📦 Installation

🧩 Community Nodes (recommended)

You can install this node directly from the n8n Community Nodes interface:

1. Go to Settings → Community Nodes in your n8n instance.
2. Enable Community Nodes if you haven’t already.
3. Enter the package name:
4. Confirm and install.

⚙️ Usage

1. Add the Reranker vLLM node to your workflow.
2. Connect it after a retriever or vector search node (e.g., Qdrant, Pinecone, Weaviate, etc.).
3. Provide:

• Query text – the user query or search question.
• Documents array – the list of retrieved text chunks.

4. Choose a supported Qwen3 Reranker model.
5. The node outputs documents reordered by their relevance scores (0–1 scale).

Reference:

• https://medium.com/@kimdoil1211/deploying-qwen3-reranker-8b-with-vllm-instruction-aware-reranking-for-next-generation-retrieval-c35a57c9f0a6

🤝 Contributing

Contributions and new model requests are welcome!
If you’d like to see support for another reranker model, please:

• Open a GitHub issue, or
• Submit a pull request with your proposed model configuration.