Elasticsearch native vector database now supports bit vectors, SIMD acceleration, and int4 quantization. The new semantic_text field type makes generating embeddings simple. ES|QL has new query capabilities and an expanded UI in Kibana Discover.

Elastic Platform 8.14 includes ES|QL GA, the use of AI for pattern recognition in logs, API key based security model for remote clusters, encryption at rest with KMS keys, retrievers, several vector optimizations, and vector quantization by default.

Elasticsearch and Kibana 8.13 deliver simplified kNN, improved query parallelization plus improvements to ES|QL, anomaly detection, and new improved health indicators. Also announcing the Elastic Integration Filter and GA of Kafka for Elastic Agent.

The 8.12 release includes improved vector data handling with efficient search, query parallelization, enhanced pipeline features, remote search status, new maintenance windows, and ES|QL query editing in Dashboard.

Elastic Stack 8.11 introduces an advanced query language known as ES|QL in the Discover application, making data exploration more straightforward and user friendly. The Elastic Learned Sparse EncodeR (ELSER) is now generally available.

Simplify configuring cross-cluster search, execute vector search faster, detect data drifts and log rate dips, stream Elastic Agent to Kafka, and authenticate Webhook connector using third-party security certificates with Elastic Stack 8.10.

Elastic Search 8.9 brings improvements to vector search and ingestion and presents hybrid search with RRF to combine vector, keyword, and semantic techniques. Public-facing search endpoints for indices are now available with search applications beta.

Elastic Stack 8.9 delivers significant speed-up of aggregations for metrics, faster and more reliable cross-cluster searches, and management of alert rules using Terraform.

In 7.11, we’re excited to announce schema on read in the Elastic Stack. We now offer the best of both worlds on a single platform. Schema on read for flexibility and schema on write for performance. We call our implementation runtime fields.