Elastic Announces Faster Filtered Vector Search with ACORN-1 and Default Better Binary Quantization Compression

New capabilities deliver up to 5X faster filtered vector search, improved ranking quality, and lower infrastructure costs to unlock scalable, cost-efficient AI applications

SAN FRANCISCO--(BUSINESS WIRE)--Elastic (NYSE: ESTC), the Search AI Company, announced new performance and cost-efficiency breakthroughs with two significant enhancements to its vector search. Users now benefit from ACORN, a smart filtering algorithm, in addition to Better Binary Quantization (BBQ) as the default for high-dimensional dense vectors. These capabilities improve both query performance and ranking quality, providing developers with new tools to build scalable, high-performance AI applications while lowering infrastructure costs.

“We’re committed to giving developers the best tools to build and iterate AI applications at scale,” said Ajay Nair, general manager, Platform at Elastic. “ACORN for filtered vector queries and default Better Binary Quantization represent a step-change in performance and efficiency. This enables our users to execute complex, high-speed, filtered queries at low latency with a dramatic memory reduction, all while maintaining high ranking quality.”

Smarter, Faster Filtered Search with ACORN

ACORN-1 is a new algorithm for filtered k-Nearest Neighbor (kNN) search in Elasticsearch. It tightly integrates filtering into the traversal of the HNSW graph, the core of Elasticsearch’s approximate nearest neighbor search engine. Unlike traditional approaches that apply filters post-search or require pre-indexing, ACORN enables flexible filter definition at query time, even after documents have been ingested.

In real-world filtered vector search benchmarks, ACORN delivers up to 5X speedups, improving latency without compromising result accuracy.

Improved Ranking with BBQ by Default

Better Binary Quantization (BBQ) is now the default quantization method for dense vectors of 384+ dimensions in Elasticsearch 9.1. This change boosts ranking quality while slashing latency and resource usage.

When benchmarked across 10 industry-standard BEIR datasets, BBQ outperformed traditional float32-based search in 9 out of 10 cases, using the NDCG@10 (Normalized Discounted Cumulative Gain at 10) metric for top-10 ranking accuracy. BBQ achieves this through a combination of aggressive compression (~32X) and evaluation of more candidates during search.

For details on how to get started with ACORN and BBQ by default in Elasticsearch 9.1, read the Elastic blog.

For more information:

• kNN search Documentation
• Vector Quantization Documentation
• Elastic Ranking Evaluation API

About Elastic

Elastic (NYSE: ESTC), the Search AI Company, integrates its deep expertise in search technology with artificial intelligence to help everyone transform all of their data into answers, actions, and outcomes. Elastic's Search AI Platform — the foundation for its search, observability, and security solutions — is used by thousands of companies, including more than 50% of the Fortune 500. Learn more at elastic.co.

Elastic and associated marks are trademarks or registered trademarks of Elasticsearch BV and its subsidiaries. All other company and product names may be trademarks of their respective owners.