OpenSearch 3.6 LTS
OpenSearch 3.6 was released as the project’s first long‑term support version with 18 months of support, SBOM compliance and a new observability stack. The update adds APM and agent tools for search, tracing and relevance tuning to support retrieval and relevance in AI systems. (techzine.eu), (cmotech.news)
OpenSearch 3.6 is now the project’s first long-term support release, giving users a version with 18 months of support instead of the usual faster upgrade cadence. (opensearch.org) OpenSearch is open-source software for storing logs, tracing requests across services, and searching large datasets; companies also use it as the retrieval layer that fetches documents for artificial intelligence systems. Version 3.6 was announced on April 15, 2026, alongside a new long-term support program from the OpenSearch Software Foundation. (opensearch.org) The foundation said the first long-term support versions are OpenSearch 3.6 and OpenSearch 2.19. It said each designated long-term support release will get at least 18 months of maintenance, software bill of materials coverage across the project, and support from accredited vendors including BigData Boutique, Eliatra, Graylog and Aiven. (opensearch.org) Version 3.6 adds a preconfigured observability stack that can be launched with one command, plus Application Performance Monitoring, which tracks how requests move through distributed applications and where delays start. The release also adds service dependency maps, performance metrics and trace views built around OpenTelemetry conventions. (opensearch.org, docs.opensearch.org, github.com) The same release pushes OpenSearch further into search for artificial intelligence systems. The project said 3.6 adds agent-guided workflows for building search applications, natural-language relevance tuning and Agent-v2 features including token-usage tracking and observability traces. (opensearch.org, docs.opensearch.org, github.com) That combination targets two jobs teams often run side by side: finding the right data for a model to use, and watching the application after it goes live. OpenSearch said the new tooling is meant for enterprise search, observability, analytics and vector database workloads under one project umbrella. (opensearch.org) The long-term support push also answers a procurement problem, not just a technical one. The foundation said its software bill of materials program scans all 150 OpenSearch repositories so enterprises can document provenance and security posture for audits and internal reviews. (opensearch.org) OpenSearch has been moving on two tracks since its 3.0 generation: faster feature releases for users who want new capabilities, and now designated long-term support versions for users who want fewer migrations. The public version history lists 3.6.0 as the release that introduced Agent-v2 updates, Application Performance Monitoring and generally available pull-based ingestion improvements. (docs.opensearch.org, github.com) For users deciding whether to upgrade, 3.6 is less about a single feature than about picking a steadier branch. The project is packaging search, tracing and relevance tuning into one supported release and telling enterprises they can stay there through an 18-month window. (opensearch.org, opensearch.org)