Elasticsearch Vector Database: Ship in minutes, scale affordably to hundreds of billions

The hard parts of hybrid retrieval, already done. Optimized defaults, third-party and native Jina AI models, and managed GPU inference are all available out of the box. Build fast, scalable AI apps, not infrastructure.

Start free trial
See documentation

What should your vector database do for you?

  • Vector search: Context, intent, relationships

    Semantic similarity search can return the right results — even when the words don’t quite match.

    Hands-on tutorial

  • Hybrid search: Precision + flexibility

    Keyword search is precision. Vector search is nuanced. Hybrid search brings both together.

    Hybrid search examples

  • Sparse vs. dense vectors: Fast and efficient

    Sparse text expansion and dense meaning matches are perfect for open-ended, real-world search.

    Document ranking types

  • Filters, ranking, reranking: Relevance with context

    Filters reduce scope, ranking finds the signal — both hard problems, but pure delight when done right.

    Filtered kNN search

Elasticsearch Vector Database: More than just vectors, loved by developers

No gaps or compromises. Optimized defaults. Hybrid retrieval. Everything works together, because it was built this way.

  • Hybrid search that understands everything

    Blend lexical search, semantic search (with third-party or native Jina AI models ), metadata, and filters in a single API call with hybrid search. Rank results by meaning, precision, and context.

  • Optimized out of the box

    Get efficient storage and indexing with vectordb_document index mode, and up to 95% memory savings with Better Binary Quantization (BBQ). Both on by default, production-ready from your first query.

  • Native GPU inference with Jina AI models

    Run embeddings and reranking on Jina AI models via Elastic Inference Service (EIS), with semantic_text handling chunking automatically. Bring your own through the Inference API.

  • More vectors, less memory, no tradeoffs

    Achieve low-memory retrieval at scale with DiskBBQ by default, while auto-calibration tunes compression and oversampling to your data automatically. Hold recall high with vector search that stays tuned as your data grows.

  • Vector-specific pricing

    Affordable at scale and predictable pricing from the start. Forecast your bill up front.

  • Enterprise from day one

    Security, RBAC, and observability are built in. Cross-project search allows customers to grow into full Elasticsearch.

One call, that's all: Zero to vector search in minutes

Get started with Elasticsearch Vector Database. The fastest path to vector search with defaults for vector workloads, hybrid search, and your choice of generating embeddings or bringing your own.

from elasticsearch import Elasticsearch

client = Elasticsearch(
   "https://xyz.elastic.cloud:443",
   api_key="..",

)

client.indices.create(
   index="my_semantic_vectors",
   mappings={
       "properties": {
           "content": {"type": "text", "copy_to": "semantic_content"},
           "semantic_content": {"type": "semantic_text"},
       }
   },
)

client.bulk(
   operations=[
       {"index": {"_index": "my_semantic_vectors"}},

       {"content": "Yellowstone National Park spans Wyoming, Montana, and Idaho, covering over 2.2 million acres. It is famous for the geyser Old Faithful and sits atop the Yellowstone Caldera, a supervolcano."},

       {"index": {"_index": "my_semantic_vectors"}},

       {"content": "Yosemite National Park covers over 750,000 acres in California. A UNESCO World Heritage Site, it is best known for its granite cliffs, waterfalls, and giant sequoia trees."},

       {"index": {"_index": "my_semantic_vectors"}},

       {"content": "Rocky Mountain National Park is known for its mountainous terrain, including Longs Peak, the highest in the park. It is a popular destination for hiking, camping, and wildlife viewing."},
   ],
)

Build with Elasticsearch Vector Database

  • Semantic search and conversational AI

    Understand intent. Search naturally. Get precise results.

    Learn more

  • Ecommerce and retail

    Use hybrid search for better discovery, ranking, and personalization.

    Learn more

  • Knowledge discovery (healthcare and finance)

    Bring documents, tickets, and emails together for semantic discovery.

    Learn more

  • Investigation workflows (public sector)

    Uncover patterns and insights across data, securely.

    Learn more

  • Use geo-distance queries for precise, real-time nearby results.

    Learn more

  • Infrastructure and database offloading

    Separate retrieval from storage for faster, scalable querying.

    Learn more

Vector DB, and the rest of what you actually need

Real-world vector search needs more than vectors. Combine semantic search with filters, security, structured fields, and BM25 ranking in a single query, no stitching required.

A high-quality neighborhood

From prompt to product, how teams are building with Elasticsearch Vector Database.

  • Customer spotlight

    FRAIM uses Elasticsearch to build a knowledge search platform for the AI era, natively consolidating text and vector search to cut its overall search platform costs by over 50% and significantly accelerate the development of RAG and AI agents.

    Learn more

  • Customer spotlight

    WP Engine uses Elasticsearch to bring AI innovation to WordPress websites, leveraging native Google Cloud integrations to reduce AI feature development time from weeks to hours and deliver 5ms search response times with zero downtime.

    Learn more

  • Customer spotlight

    Adobe scales, manages multiple use cases, and puts machine learning features to work with Elastic.

    Learn more

Results, by the numbers

Elasticsearch Vector Database performance benchmarks and results.

Get started: Resources for every stage of the build

Ingest, parse, and index
Embed
Retrieve
Dev tools
Flexible document model
Store embeddings (HNSW for speed, BBQ for compression, DiskBBQ for scale)
Search embeddings (vector search with third-party or SOTA multilingual and multimodal Jina models)
Piped queries ES|QL: Piped query language for filtering, transforming, and analyzing data
Secure storage (document- and field-level security)
HNSW: Fast approximate kNN search optimized for high recall and low latency at scale
Full text search (BM25)
Explore and visualize data stored in Elasticsearch with Kibana
Process structured and unstructured data
BBQ: 32x vector compression that preserves search relevance while slashing memory and compute costs
Native hybrid search (BM25 + vector search)
Create search interfaces and configurable Search UI components with just a few lines of code
Ingest tools (clients, web crawler,* connectors,* inference pipelines*)
DiskBBQ: Disk-based vector search that cuts RAM requirements for large-scale, memory-constrained environments
Native Jina reranker models and Learning-to-Rank
Access AI-powered capabilities for developing and interacting with agents that work with your Elasticsearch data with Agent Builder
Real-time document and metadata updates
vectordb_document mode: Applies best-practice vector defaults automatically (quantization method, merge policy, cache loading) for an optimal setup out of the box
Filtering, faceting, aggregations

Search autocomplete

semantic_text: Automatically generate embeddings and handle chunking, and add semantic search with no pipeline setup
Auto-calibration: Samples your stored vectors to pick the best compression and oversampling, adapting on the fly as your data changes
Optimized for multiple data types (text, vector, geo, and more)
Ingest, parse, and index
Flexible document model
Secure storage (document- and field-level security)
Process structured and unstructured data
Ingest tools (clients, web crawler,* connectors,* inference pipelines*)
Real-time document and metadata updates
semantic_text: Automatically generate embeddings and handle chunking, and add semantic search with no pipeline setup
Embed
Retrieve
Dev tools
Store embeddings (HNSW for speed, BBQ for compression, DiskBBQ for scale)
Search embeddings (vector search with third-party or SOTA multilingual and multimodal Jina models)
Piped queries ES|QL: Piped query language for filtering, transforming, and analyzing data
HNSW: Fast approximate kNN search optimized for high recall and low latency at scale
Full text search (BM25)
Explore and visualize data stored in Elasticsearch with Kibana
BBQ: 32x vector compression that preserves search relevance while slashing memory and compute costs
Native hybrid search (BM25 + vector search)
Create search interfaces and configurable Search UI components with just a few lines of code
DiskBBQ: Disk-based vector search that cuts RAM requirements for large-scale, memory-constrained environments
Native Jina reranker models and Learning-to-Rank
Access AI-powered capabilities for developing and interacting with agents that work with your Elasticsearch data with Agent Builder
vectordb_document mode: Applies best-practice vector defaults automatically (quantization method, merge policy, cache loading) for an optimal setup out of the box
Filtering, faceting, aggregations

Search autocomplete

Auto-calibration: Samples your stored vectors to pick the best compression and oversampling, adapting on the fly as your data changes
Optimized for multiple data types (text, vector, geo, and more)

Frequently asked questions

What is a vector database and how does it work?

A vector database stores information as vectors, which are numerical representations of data objects, also known as vector embeddings. It uses vector embeddings for multimodal search across a massive dataset of structured, unstructured, and semi-structured data, such as images, text, videos, and audio. Vector databases are built to manage vector embeddings and therefore offer a complete solution for data management.

What are vector embeddings?

Vector embeddings use a machine learning model to translate text into numbers, allowing you to perform vector searches. By converting data into vectors, embeddings make it easier to compare, search, and analyze similarities between items in this space.

What are the benefits of a vector database?

A vector database offers efficiency at scale by enabling seamless data migration across on-premises, air-gapped, and sovereign cloud environments and providing storage for vector embeddings.

Vector databases excel at similarity search, allowing you to find related items easily, which is essential for recommendation systems, image search, and content discovery. With semantic search capabilities, they go beyond simple keyword matching to deliver results based on meaning and context. By storing vector embeddings, they support AI and machine learning applications, making it easier to deploy natural language processing (NLP) and recommendation models.

Is Elasticsearch a vector database?

Yes, Elasticsearch is the world's most widely deployed, open source vector database, offering you an efficient way to create, store, and search vector embeddings at scale. With Elastic's enterprise-ready vector database, you achieve fast query times and optimal performance, even with rapidly changing data. Built to scale, it delivers relevant, personalized search results while simplifying development processes.

Why choose Elastic as your vector database?

Elasticsearch is the world's most downloaded vector database, purpose-built to help developers deliver high-precision search and scalable AI experiences faster and at lower costs. Elasticsearch Vector Database handles the hard parts of hybrid retrieval with optimized defaults, native Jina AI and third-party models, and managed GPU inference, all out of the box. Run vector and keyword search across text, image, and multimodal data on a single index, with no embedding pipeline to build or maintain. It ships pre-tuned for vector workloads, delivering high-performance, absorbing traffic spikes automatically and scaling back when demand drops. Get the security, RBAC, and recovery paths you expect, so you build apps, not infrastructure.

Can I run Elasticsearch as an on-premises or air-gapped vector database?

Yes. Elasticsearch is fully deployable on-premises — on bare metal, in a private cloud, or in completely air-gapped networks with no external connectivity. Government agencies, defense contractors, and regulated enterprises use Elastic Cloud Enterprise (ECE) to orchestrate on-premises Elasticsearch clusters at scale, including in classified and disconnected environments. All vector search, hybrid search, and RAG capabilities available on Elastic Cloud are equally available in on-premises deployments.