OpenSearch

What is OpenSearch?

OpenSearch is an open-source, Apache 2.0-licensed search and analytics suite built on Apache Lucene. It was forked from Elasticsearch in 2021 after Elastic's license change, and is now governed by the OpenSearch Software Foundation under the Linux Foundation (with major backing from AWS). It combines three things in one system: full-text (BM25) keyword search, vector search through its k-NN vector engine, and a built-in observability and dashboards stack for logs and metrics.

Vector and AI search

For AI workloads, OpenSearch's vector engine handles k-nearest-neighbor similarity search and supports semantic, hybrid (BM25 plus vectors), multimodal, and neural sparse search, along with RAG-style conversation search. It integrates with common LLM frameworks as a vector store and provides ingest pipelines for generating and transforming embeddings. That breadth is the reason it shows up both as a search engine and as a vector database option.

Pricing

OpenSearch itself is free and open-source. You can self-host it on your own infrastructure at no license cost, or run it as a managed service, Amazon OpenSearch Service is the best known, with other providers also offering hosted OpenSearch. The cost in practice is the compute and the operational burden of running a distributed cluster, or the managed-service fee if you offload that.

Who it's for

OpenSearch suits teams that want one system for both keyword/log search and vector retrieval, especially those already on the Elasticsearch/Lucene model or running observability workloads who want to add semantic search without a separate vector database. Compared with a purpose-built vector database (Qdrant, Weaviate, Pinecone) it carries more operational weight but covers far more ground; compared with Elasticsearch it is the fully open-source path. For a lighter open-source search engine, Meilisearch and Typesense are the simpler alternatives.