Sold by: MongoDB, Inc
Deployed on AWS
Free Trial
General-purpose reranker with instruction following optimized for both latency and quality, with improved long-context and code retrieval. Context length: 32K for queries & documents (8K for queries).
Overview
Rerankers are neural networks that predict the relevancy scores between a query and documents and rank them based on the scores. They are used to refine search results in semantic search/retrieval systems and retrieval-augmented generation (RAG).
rerank-3-lite is the next generation of Voyage AI's reranker optimized for both latency and quality, and a drop-in upgrade to rerank-2.5-lite that requires no code changes. Trained with an updated backbone and an improved mixture of training data, rerank-3-lite improves on rerank-2.5-lite across nearly all domains, with the largest gains on long documents and code, and is roughly on par with the retrieval quality of rerank-2.5. Relevance scores are calibrated to match the score distribution of rerank-2.5-lite, so score thresholds tuned on rerank-2.5-lite continue to work.
Averaged across 95 retrieval datasets in 9 domains and four first-stage retrieval methods, rerank-3-lite outperforms Cohere Rerank v4.0 Pro, Qwen3-Reranker-8B, and rerank-2.5-lite by 2.23%, 2.53%, and 1.10% NDCG@10, respectively - outperforming Qwen3-Reranker-8B, the leading open-weights reranker, despite being over an order of magnitude smaller. On long-document retrieval, rerank-3-lite outperforms rerank-2.5-lite by 1.86% and Cohere Rerank v4.0 Pro by 13.80%. On code retrieval, it outperforms rerank-2.5-lite by more than 3.5% atop both voyage-3-large and voyage-4-large.
The model supports a combined context length of 32K tokens per query-document pair, including up to 8K tokens for the query, enabling more accurate retrieval over longer documents. rerank-3-lite also retains the instruction-following capability introduced in the rerank-2.5 series, allowing users to guide relevance scoring through natural language instructions. On the MAIR instruction-following benchmark, it outperforms rerank-2.5-lite by 0.91% and Cohere Rerank v4.0 Pro by 2.34%.
Learn more about rerank-3-lite here: https://blog.voyageai.com/2026/09/30/rerank-3
Highlights
- Optimized for both latency and quality, and a drop-in upgrade to rerank-2.5-lite with no code changes: same API, 32K-token combined context length per query-document pair (up to 8K for the query), and instruction following, with relevance scores calibrated so existing rerank-2.5-lite score thresholds continue to work.
- Outperforms Cohere Rerank v4.0 Pro by 2.23%, Qwen3-Reranker-8B by 2.53%, and rerank-2.5-lite by 1.10% NDCG@10 on average across 95 retrieval datasets spanning 9 domains, and is roughly on par with the retrieval quality of rerank-2.5.
- Largest gains on long documents and code: outperforms rerank-2.5-lite by 1.86% and Cohere Rerank v4.0 Pro by 13.80% on long-document retrieval, and rerank-2.5-lite by more than 3.5% on code retrieval atop both voyage-3-large and voyage-4-large.
Details
Sold by
Delivery method
Latest version
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
Free trial
Try this product free for 7 days according to the free trial terms set by the vendor.
Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimension | Description | Cost/host/hour |
|---|---|---|
ml.g6.xlarge Inference (Real-Time) Recommended | Model inference on the ml.g6.xlarge instance type, real-time mode | $2.25 |
ml.g5.xlarge Inference (Batch) Recommended | Model inference on the ml.g5.xlarge instance type, batch mode | $35.92 |
ml.g5.xlarge Inference (Real-Time) | Model inference on the ml.g5.xlarge instance type, real-time mode | $2.82 |
ml.g5.2xlarge Inference (Real-Time) | Model inference on the ml.g5.2xlarge instance type, real-time mode | $3.03 |
ml.g5.4xlarge Inference (Real-Time) | Model inference on the ml.g5.4xlarge instance type, real-time mode | $4.06 |
ml.g5.8xlarge Inference (Real-Time) | Model inference on the ml.g5.8xlarge instance type, real-time mode | $6.12 |
ml.g6.2xlarge Inference (Real-Time) | Model inference on the ml.g6.2xlarge instance type, real-time mode | $2.44 |
ml.g6.4xlarge Inference (Real-Time) | Model inference on the ml.g6.4xlarge instance type, real-time mode | $3.31 |
ml.g6.8xlarge Inference (Real-Time) | Model inference on the ml.g6.8xlarge instance type, real-time mode | $5.04 |
ml.g7e.2xlarge Inference (Real-Time) | Model inference on the ml.g7e.2xlarge instance type, real-time mode | $4.49 |
Dimensions summary
You pay by the host hour based on the AWS instance you run the reranker model on. Seventeen options cover real-time inference across GPU instance families in a range of sizes, from smaller single-GPU types to larger multi-GPU types. One option covers batch inference on the ml.g5.xlarge instance. Pricing scales with the instance size and family you choose, plus the inference mode. Real-time endpoints run continuously, while batch mode runs finite jobs on datasets. Your total hourly cost combines this software rate with the underlying AWS infrastructure rate.
Top-of-mind questions for buyers
One host hour is one hour that a deployed instance runs as a persistent inference endpoint. Real-time endpoints run continuously and bill per running hour until you delete them. Canceling your subscription does not stop active endpoints, so delete endpoints you no longer need to avoid charges.
Real-time endpoints stay running and bill per host hour for request-by-request inference. The batch option on ml.g5.xlarge runs a finite job that processes a dataset and writes predictions to a file, then ends. Batch suits bulk processing; real-time suits ongoing live requests.
No. Your total hourly cost combines two parts. The software rate shown here covers model usage. The underlying AWS instance infrastructure is billed separately at its own hourly rate. Rates also vary by instance type and region. Add both to estimate your true hourly spend.
docs.voyageai.com
Helpful?
Vendor refund policy
Refunds are processed according to the EULA. For assistance, contact aws-marketplace@mongodb.com
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Deploy the model on Amazon SageMaker AI using the following options:
Version release notes
MongoDB is excited to announce the initial release of rerank-3-lite.
Additional details
Inputs
- Summary
Supply a query and a list of documents to score for relevance.
Note: Does NOT support batch transform.
- Limitations for input type
- Total request size, calculated as the query tokens multiplied by the number of documents plus the tokens of all documents, cannot exceed 600,000 tokens.
- Input MIME type
- application/json
https://github.com/voyage-ai/voyageai-aws/blob/main/sample_reranker_input.json
rerank-3-lite reranker model does NOT support batch transform.
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
query | A query string (string). | Max 8,000 tokens. | Yes |
documents | Documents to rerank (List[string]). | Maximum of 1,000 documents; max 32,000 total tokens for each document. | Yes |
top_k | Number of most relevant documents to return (int). | Default: null (all documents are returned). | No |
truncation | Whether to truncate inputs to fit context limits (boolean). | Default: true. | No |
return_documents | Whether to return the documents in the response (boolean). | Default: false. | No |
Resources
Vendor resources
Support
Vendor support
Please email us at aws-marketplace@mongodb.com for inquiries and customer support.
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
Rerank will return a sorted list of documents based on the semantic similarity between the query and documents.
General-purpose reranker with instruction following optimized for both latency and quality. Context length: 32K for queries & documents (8K for queries).
Rerank will return a sorted list of documents based on the semantic similarity between the query and documents.
Rerank will return a sorted list of documents based on the semantic similarity between the query and documents in over 100 languages
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.