Elastic

license

AGPL-3.0(OSI)

enterprise-dir

Elastic-License-2.0

source

public

core features withheld

yes — the x-pack directory is licensed solely under the Elastic License 2.0 rather than the repository's default triple license, so a set of features in the published tree cannot be run under the open terms

The repository's LICENSE sets a default triple license — AGPL-3.0, SSPL-1.0 or Elastic License 2.0, at the recipient's choice — so an OSI-approved option covers most of the tree. The x-pack directory, though, is licensed solely under Elastic License 2.0, which makes Elasticsearch open core rather than fully open source.

• https://raw.githubusercontent.com/elastic/elasticsearch/main/LICENSE.txt recorded 2026-09-30

  LICENSE.txt sets a default triple license and says code licensed solely under Elastic License 2.0 "is found only in the x-pack folder", while "The default throughout the repository is a triple" license.

• https://raw.githubusercontent.com/elastic/elasticsearch/main/README.asciidoc recorded 2026-09-30

  README calls Elasticsearch "a distributed search and analytics engine, scalable data store and vector database" and says the local trial license "includes all Elastic features" before reverting to "Free and open - Basic"; searched the body for enterprise, paid tier and license key: none named.

• https://raw.githubusercontent.com/elastic/elasticsearch/main/x-pack/plugin/inference/build.gradle recorded 2026-09-30

  Inference plugin build file under x-pack carries a header reading "Licensed under the Elastic License" 2.0.

• https://raw.githubusercontent.com/elastic/elasticsearch/main/x-pack/plugin/ml/build.gradle recorded 2026-09-30

  ML plugin build file under x-pack carries a header reading "Licensed under the Elastic License" 2.0.

Elasticsearch's adoption reading is based on GitHub stars, since Elastic publishes no server deployment figure and the `elasticsearch` package on PyPI is the Python client rather than the server. Client downloads measure only client installs, and a star does not confirm production use, so this reading cannot show how widely the server itself is actually deployed.

• https://api.github.com/repos/elastic/elasticsearch recorded 2026-08-18

  Repo metadata - stargazers_count = 77,824 - for elastic/elasticsearch.

Elasticsearch hosts and evaluates its own ranking and embedding models inside the serving path through its in-tree inference and ml plugins, as Vespa does. That sets it apart from vector databases, which fuse or filter retrieval scores but do not host a model themselves.

• https://raw.githubusercontent.com/elastic/elasticsearch/main/README.asciidoc recorded 2026-09-30

  README lists "Retrieval Augmented Generation (RAG)", vector search and full-text search among use cases of "a distributed search and analytics engine, scalable data store and vector database".

• https://www.elastic.co/docs/deploy-manage/distributed-architecture/clusters-nodes-shards recorded 2026-09-30

  Cluster docs state "Each document in an index belongs to one primary shard." and "A replica shard is a copy of a primary shard."

• https://www.elastic.co/docs/explore-analyze/elastic-inference/inference-api recorded 2026-09-30

  Inference API docs show an endpoint using the elasticsearch service that deploys the ELSER model, where "The API call will automatically download and deploy the model" for use in semantic_text fields.

• https://www.elastic.co/docs/solutions/search/ranking/semantic-reranking recorded 2026-09-30

  Semantic reranking page says "Both use the same underlying inference endpoints and re-ranking models." for the text_similarity_reranker retriever and the ES|QL RERANK command.

• https://www.elastic.co/docs/solutions/search/vector recorded 2026-09-30

  Vector search docs list dense_vector and sparse_vector fields, the semantic_text field type with "automatic embedding generation and model management", and hybrid search tutorials.