Elastic, the Search AI Company, has announced a significant leap in vector search performance and efficiency with the introduction of two powerful innovations: the ACORN-1 algorithm and Better Binary Quantization (BBQ). These enhancements aim to help developers build scalable, high-performance AI applications while dramatically reducing latency and infrastructure expenses.
At the core of these improvements is ACORN-1, a smart filtering algorithm designed specifically for filtered k-Nearest Neighbor (kNN) search in Elasticsearch. Unlike traditional methods that apply filters after search results are generated or require pre-indexing strategies, ACORN integrates the filtering process directly into the traversal of the HNSW (Hierarchical Navigable Small World) graph — the backbone of Elasticsearch’s approximate nearest neighbor engine. This design allows for flexible, on-the-fly filtering even after document ingestion, enabling real-time adaptability with minimal trade-offs. Benchmark results show up to 5X speedups in filtered search queries without compromising accuracy, offering game-changing performance for AI workloads requiring dynamic filtering.
Complementing ACORN-1, Elastic has made Better Binary Quantization (BBQ) the default quantization method for dense vectors with 384 or more dimensions in Elasticsearch 9.1. BBQ improves query performance and ranking quality by compressing vectors up to 32 times while evaluating a broader set of candidates during search. In benchmarking tests across 10 widely used BEIR datasets, BBQ outperformed traditional float32-based search in 9 out of 10 cases using the NDCG@10 metric, showcasing superior top-10 ranking accuracy.
“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.”
Together, ACORN and BBQ position Elastic as a leader in delivering cost-effective, high-speed AI search capabilities, particularly for use cases in semantic search, recommendation engines, and intelligent retrieval systems. These enhancements align with Elastic’s broader mission to enable enterprises to harness the full potential of AI-powered search while maintaining speed, flexibility, and control at scale.
