How to Intermediate · 3 min read

How to use reranker in LlamaIndex

Q: How to use reranker in LlamaIndex

Use the Reranker class in LlamaIndex to improve search result relevance by reordering retrieved documents based on a reranking model. Initialize a reranker with a supported model, then pass it to your Retriever or QueryEngine to apply reranking on search results.

Quick answer

Use the Reranker class in LlamaIndex to improve search result relevance by reordering retrieved documents based on a reranking model. Initialize a reranker with a supported model, then pass it to your Retriever or QueryEngine to apply reranking on search results.

PREREQUISITES

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

Setup

Install llama-index and openai packages, and set your OpenAI API key as an environment variable.

bash

pip install llama-index openai>=1.0

Step by step

This example shows how to create a Reranker using an OpenAI model and integrate it with a Retriever to rerank search results in LlamaIndex.

python

import os
from llama_index import SimpleDirectoryReader, GPTVectorStoreIndex, ServiceContext, Reranker
from llama_index.retrievers import VectorIndexRetriever
from openai import OpenAI

# Set up OpenAI client
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Load documents from a directory
documents = SimpleDirectoryReader("./data").load_data()

# Create an index
index = GPTVectorStoreIndex.from_documents(documents)

# Initialize a reranker with an OpenAI model
reranker = Reranker.from_openai(
    client=client,
    model="gpt-4o",
    max_rerank=5  # number of top docs to rerank
)

# Create a retriever with reranker
retriever = VectorIndexRetriever(
    index=index,
    reranker=reranker
)

# Query with reranking
query = "Explain the benefits of renewable energy."
results = retriever.retrieve(query)

# Print reranked results
for i, doc in enumerate(results):
    print(f"Result {i+1}: {doc.get_text()[:200]}...\n")

output

Result 1: Renewable energy offers sustainable power sources that reduce greenhouse gas emissions and dependence on fossil fuels...

Result 2: Using renewable energy can lower energy costs over time and create jobs in new industries...

Result 3: It improves air quality and public health by reducing pollution from traditional energy sources...

Result 4: Renewable energy sources are abundant and can help stabilize energy prices...

Result 5: Adoption of renewable energy supports energy independence and national security...

Common variations

Use different models for reranking, such as gpt-4o-mini or other supported OpenAI models.
Apply reranking in an async context by adapting the retriever calls accordingly.
Combine reranking with other retriever types like BM25Retriever or custom retrievers.

Troubleshooting

If reranking results do not improve relevance, verify your reranker model and max_rerank parameter.
Ensure your OpenAI API key is correctly set in os.environ["OPENAI_API_KEY"].
Check for network issues or API rate limits if requests fail.

✅

Key Takeaways

Use Reranker with a supported model to reorder search results for better relevance.
Integrate the reranker into your Retriever to apply reranking seamlessly.
Adjust max_rerank to control how many top documents get reranked for efficiency.
Always set your API key securely via environment variables to avoid credential leaks.
Test different models and retriever combinations to optimize search quality.

Verified 2026-04 · gpt-4o

Verify ↗