For AI agents: this page is also available as Markdown at /docs/roadmap-and-status.md. Every docs page is: add .md to its URL. llms.txt lists them all.

Feature status

This page lists the capabilities people ask about most, with their status. Available means you can use it now. In development means the work is under way but not released, and we give no dates. Not supported means it is not offered today and is not in progress.

Search and relevance

Capability Status Notes
Keyword search with BM25 Available Searching
Phrase, fuzzy, prefix, wildcard and regex queries Available Wildcard and regex depend on your plan
Filters, sorting, pagination Available Pagination is offset based, see below
Facets, range and date histograms, nested facets Available Facets and aggregations
Geo box and distance queries Available Polygon queries are in development
Minimum score cut-offs Available
Result collapsing and diversification Available
Highlighting and snippets In development
Cardinality and percentile metrics Available In /v1 metrics and Elasticsearch-compatible aggregations
Cursor pagination, scroll, search_after, point in time Not supported Page with offset and limit, or narrow with filters
Suggesters, autocomplete, more-like-this Not supported
Synonyms, stop words, language analyzers other than English stemming Not supported N-gram and shingle filters are in development
Capability Status Notes
Dense vectors with cosine, inner product and L2 Available Vector search
Exact search and approximate search Available IVF with PQ quantization. Per-field codec settings (ann) are not exposed in the API yet
Filtered vector search Available Filters apply inside the search
Matryoshka embeddings Available with client-side truncation Server-side truncate_dims is implemented in the engine and not yet exposed in the API, Matryoshka embeddings
Multiple vectors per document In development for the public API Implemented in the engine, scored by the best vector, Multi-vector fields
Late interaction (summed MaxSim) scoring In development
Hybrid search with a knn section beside a query Available Scores are summed, Hybrid search
RRF and weighted linear fusion In development for the public API Implemented in the engine
float16 and int8 vector storage In development Vectors are stored as float32 today
Sparse vectors and rank features In development
Binary or bit vectors Not supported
HNSW graph indexes Not supported
Automatic embedding at ingest or query time Not supported Send vectors you computed yourself
Rerankers and cross-encoders Not supported Rerank the top results in your application

Data and schema

Capability Status Notes
Text, string, int, float, double, date, vector and geo point fields Available Schemas and field types
Boolean, IP and decimal fields In development Booleans are stored as text today
Flattened objects and range field types In development
Nested and join documents Not supported Flatten, or model relations with ids
Replace a whole document by id Available
Partial document updates, delete by query Not supported
Live schema changes Available They apply to new documents, there is no backfill

Compatibility and integrations

Capability Status Notes
Elasticsearch API Available Elasticsearch compatibility
OpenSearch API Available OpenSearch compatibility
Solr API Available Solr compatibility
SQL, Lucene, LogQL and Solr queries through the REST API Available Query languages
Loki push and query HTTP API Not supported Loki and LogQL explains the alternatives
MCP server for AI agents Available Guide for AI agents
Kibana, OpenSearch Dashboards Not supported Grafana works through the Elasticsearch or OpenSearch data source
Single sign-on Not supported

If something you need is marked in development or not supported, tell us at /contact so we can weigh it.

View this page as Markdown