Overview

OpenSearch is a scalable, open-source search and analytics suite used for full-text search, application monitoring, log analytics and vector search. It is licensed under Apache 2.0 and developed in the open by its community. OpenSearch Dashboards provides the visualisation and management interface for the suite. The project website offers downloads of OpenSearch and access to the documentation.

  • Apache 2.0-licensed open source
  • Full-text search, application monitoring, log analytics and vector search
  • OpenSearch Dashboards for visualisation and management
  • Positioned as the open engine for AI-powered search and observability

Sources: Home - OpenSearch, OpenSearch and OpenSearch Dashboards | OpenSearch Documentation

Capabilities

The project groups its functionality into four capability areas. Machine learning and AI covers vector search and anomaly detection for building AI/ML applications. Search covers enterprise search and document search, while observability covers the observability stack and log analysis. Security analytics addresses detecting and responding to threats in real time through threat intelligence and event correlation.

  • Machine learning and AI — vector search, anomaly detection
  • Search — enterprise search, document search
  • Observability — observability stack, log analysis
  • Security analytics — threat intelligence, event correlation

Source: Home - OpenSearch

Search techniques

OpenSearch supports keyword-based lexical search using BM25, which is the default probabilistic scoring algorithm and is used in queries such as multi_match, range and term. Semantic search uses dense vector embeddings stored as vector fields and retrieved with k-nearest neighbour algorithms. Hybrid search combines lexical and semantic results, using score-based normalisation such as min-max or rank-based normalisation such as reciprocal rank fusion. Neural sparse search, introduced in OpenSearch 2.11, encodes text into (token, weight) entries so that indexes and searches can be built on Lucene's inverted index.

  • Lexical search with BM25
  • Semantic search with dense vector embeddings and k-NN
  • Hybrid search with score-based and rank-based normalisation
  • Neural sparse search with v1 and v2 sparse encoding models

Sources: Building effective hybrid search in OpenSearch: Techniques and best practices - OpenSearch, Improving search efficiency and accuracy with the newest v2 neural sparse models - OpenSearch

Vector database features

OpenSearch is used as a vector database for workloads including retrieval-augmented generation, where relevant content is retrieved from a large document collection to improve large language model responses. Approximate nearest neighbour search uses the HNSW algorithm, tuned through the M, efSearch and efConstruction hyperparameters. The k-NN plugin integrates the Faiss scalar quantizer (SQfp16), available from OpenSearch 2.13, which stores vectors at 16-bit precision and saves roughly half the memory with minimal reduction in recall. OpenSearch 2.18 introduced support for Intel AVX-512 instructions, with benchmarks showing up to 15% faster indexing and up to 13% faster search compared with AVX2.

  • HNSW approximate nearest neighbour search
  • Faiss FP16 scalar quantization (SQfp16) in the k-NN plugin
  • Intel AVX-512 acceleration from OpenSearch 2.18
  • Documented RAG setup using OpenSearch with the DeepSeek model

Sources: A practical guide to selecting HNSW hyperparameters - OpenSearch, Optimizing OpenSearch with Faiss FP16 scalar quantization: Enhancing memory efficiency and cost-effectiveness - OpenSearch, Boost OpenSearch vector search performance with Intel AVX-512 - OpenSearch, Zero to RAG: A quick OpenSearch vector database and DeepSeek integration guide - OpenSearch

Related components and tooling

The OpenSearch Platform brings together the project's integrated components. Data Prepper is an open-source data collector that collects, filters, enriches and aggregates trace and log data for downstream analysis and visualisation in OpenSearch; its pipelines consist of a source, optional processors and one or more sinks. The Anomaly Detection plugin is powered by the Random Cut Forest algorithm, which received four algorithmic enhancements in OpenSearch 2.17 aimed at reducing false positives. OpenSearch Dashboards development is documented alongside VS Code workflows covering Jest testing, linting and server run tasks.

  • OpenSearch Platform — integrated components
  • Data Prepper — trace and log data collection
  • Anomaly Detection plugin — Random Cut Forest algorithm
  • OpenSearch Dashboards — visualisation and management interface

Sources: Home - OpenSearch, Step-by-step: Creating a new database integration using Data Prepper - OpenSearch, Enhancing OpenSearch anomaly detection: Reducing false positives through algorithmic improvements - OpenSearch, Streamline OpenSearch Dashboards development with VS Code - OpenSearch

Platform security changes

OpenSearch is built on Apache Lucene and provides a REST API layer for accessing documents stored in Lucene shards, along with built-in cluster management. OpenSearch 3.0.0 replaced the Java Security Manager, which has been deprecated since JDK 17 and is scheduled for removal in JDK 24, and which is incompatible with virtual threads. The project considered several alternatives, including out-of-process extensions, GraalVM, systemd hardening, a separate Java agent and complete removal of JSM-related functionality.

  • Built on Apache Lucene with a REST API layer
  • Java Security Manager removed in OpenSearch 3.0.0
  • systemd hardening and a Java agent adopted as replacements

Source: Finding a replacement for JSM in OpenSearch 3.0 - OpenSearch

Governance and community

The OpenSearch Software Foundation was formed in September 2024 as a project of the Linux Foundation and is described as the next chapter for the OpenSearch Project. The community participates in building the project and supporting other users through several collaboration channels. The Linux Foundation Research and the OpenSearch Software Foundation jointly produced the 2026 Open Data Infrastructure Report, and the project runs an ongoing observability webinar series featuring community experts.

  • Forum
  • Events
  • Blog
  • GitHub
  • Slack

Sources: Home - OpenSearch, OpenSearch and OpenSearch Dashboards | OpenSearch Documentation

Access and documentation

OpenSearch is distributed from its website, which links to downloads and to the project documentation. The documentation site covers OpenSearch and OpenSearch Dashboards by version. Community members can also reach the project through its forum, Slack workspace, GitHub repositories and blog.

  • Website: https://opensearch.org
  • Documentation: https://docs.opensearch.org/latest/about/
  • Downloads linked from the project home page

Sources: Home - OpenSearch, OpenSearch and OpenSearch Dashboards | OpenSearch Documentation

Sources

  1. Home - OpenSearch https://opensearch.org/ Verified 01 Nov 2026
  2. OpenSearch and OpenSearch Dashboards | OpenSearch Documentation https://docs.opensearch.org/latest/about/ Verified 17 Oct 2026
  3. FAQ - OpenSearch https://opensearch.org/faq/ Verified 17 Oct 2026
  4. Platform - OpenSearch https://opensearch.org/platform/ Verified 17 Oct 2026
  5. OpenSearch Agent Skills - OpenSearch https://opensearch.org/agent-skills/ Verified 17 Oct 2026
  6. Enterprise Search - OpenSearch https://opensearch.org/enterprise-search/ Verified 17 Oct 2026
  7. Community Projects - OpenSearch https://opensearch.org/community-projects/ Verified 17 Oct 2026
  8. Authors List - OpenSearch https://opensearch.org/authors-list/ Verified 17 Oct 2026
  9. Download & Get Started - OpenSearch https://opensearch.org/downloads/ Verified 17 Oct 2026
  10. Release Schedule and Maintenance Policy - OpenSearch https://opensearch.org/releases/ Verified 17 Oct 2026
  11. Testimonials - OpenSearch https://opensearch.org/testimonials/ Verified 17 Oct 2026
  12. OpenSearch Trademark Policy - OpenSearch https://opensearch.org/trademark-brand-policy/ Verified 17 Oct 2026
  13. OpenSearch Trademark Policy - OpenSearch https://opensearch.org/trademark-brand-policy.html
  14. Finding a replacement for JSM in OpenSearch 3.0 - OpenSearch https://opensearch.org/blog/finding-a-replacement-for-jsm-in-opensearch-3-0/ Verified 01 Nov 2026
  15. Building effective hybrid search in OpenSearch: Techniques and best practices - OpenSearch https://opensearch.org/blog/building-effective-hybrid-search-in-opensearch-techniques-and-best-practices/ Verified 01 Nov 2026
  16. A practical guide to selecting HNSW hyperparameters - OpenSearch https://opensearch.org/blog/a-practical-guide-to-selecting-hnsw-hyperparameters/ Verified 01 Nov 2026
  17. Boost OpenSearch vector search performance with Intel AVX-512 - OpenSearch https://opensearch.org/blog/boost-opensearch-vector-search-performance-with-intel-avx512/ Verified 01 Nov 2026
  18. Enhancing OpenSearch anomaly detection: Reducing false positives through algorithmic improvements - OpenSearch https://opensearch.org/blog/reducing-false-positives-through-algorithmic-improvements/ Verified 01 Nov 2026
  19. Zero to RAG: A quick OpenSearch vector database and DeepSeek integration guide - OpenSearch https://opensearch.org/blog/deepseek-integration-rag/ Verified 01 Nov 2026
  20. Streamline OpenSearch Dashboards development with VS Code - OpenSearch https://opensearch.org/blog/vscode-osd-setup/ Verified 01 Nov 2026
  21. Step-by-step: Creating a new database integration using Data Prepper - OpenSearch https://opensearch.org/blog/creating-a-new-database-integration-with-data-prepper/ Verified 01 Nov 2026
  22. Improving search efficiency and accuracy with the newest v2 neural sparse models - OpenSearch https://opensearch.org/blog/neural-sparse-v2-models/ Verified 01 Nov 2026
  23. Optimizing OpenSearch with Faiss FP16 scalar quantization: Enhancing memory efficiency and cost-effectiveness - OpenSearch https://opensearch.org/blog/optimizing-opensearch-with-fp16-quantization/ Verified 01 Nov 2026

Last verified 01 Nov 2026. This entry is compiled from the public web pages listed above. Nothing here is stated that those pages do not, and each of them was read on the date shown.