Last checked and updated on April 15, 2026What Is New in Elasticsearch 8.13
Elasticsearch 8.13 delivers significant enhancements across search, analytics, and management, with a strong focus on improving the developer and operator experience. This release introduces more natural querying, better observability into system behavior, and expanded platform integrations.
| Category | Key Updates |
|---|---|
| New Features | Natural Language Querying (ELSER v2), _terms_enrich stats, Vector tile search API, Search application telemetry |
| Enhancements | Faster terms aggregations, Improved recovery prioritization, Better TSDB disk usage, Geohex grid aggregation |
| Management | Remote cluster migration assistant, Service account API keys, Synthetics private locations |
| Security | FIPS 140-2 compliance (hardware mode), JWT realm support for client authentication |
How does Elasticsearch 8.13 improve semantic search?
The headline feature is the general availability of ELSER v2, a new inference model for semantic search. This model delivers significantly better performance and accuracy out-of-the-box compared to its predecessor. You can now perform natural language queries without the complexity of managing your own embeddings.
In practice, this means you can index a document and then later query it using phrases like "How do I reset my password?" instead of relying on exact keyword matches. The model handles the semantic understanding, making search experiences feel much more intuitive for end-users.
What analytics and aggregation upgrades are included?
This release tackles performance and functionality for data analysis. The terms aggregation now uses global ordinals by default, which can drastically speed up execution times for this common operation, especially on larger datasets.
For geospatial work, the new geohex_grid aggregation allows for aggregating data into Uber's H3 hexagon cells. This is a powerful addition for building location-based analytics and visualizations. Furthermore, the vector tile search API enables efficient rendering of complex geographic data for mapping applications.
How is observability and management made easier?
New telemetry for Search Applications provides crucial insights into how your search features are performing. You get data on query latency, frequency, and result quality, which is essential for tuning relevance and performance.
For infrastructure management, the remote cluster migration assistant simplifies the often-complex process of moving to a new deployment topology. Additionally, synthetics private locations give you more control over how and where you run your synthetic monitoring checks.
What about security and compliance updates?
Elasticsearch now meets the FIPS 140-2 compliance standard when running on a FIPS-enabled JVM in hardware mode. This is critical for deployments in environments with strict federal security requirements.
The JWT realm has been extended to support client authentication, providing a more secure and standardized way for programmatic clients to authenticate against the cluster using JWTs instead of traditional API keys or basic auth.
FAQ
Is ELSER v2 backwards compatible with the experimental version?
While the underlying model is improved, the usage pattern via the inference API remains consistent. You should expect to re-index your content with the new model to benefit from its enhanced accuracy and performance, as the embeddings are different.
What's the real-world performance impact of the new terms aggregation?
The shift to using global ordinals by default can reduce memory pressure and improve query latency for aggregations that were previously building ordinals on-the-fly. The gains are most noticeable on large, immutable indices where ordinals can be reused effectively.
When should I use the geohex_grid aggregation over the standard geohash_grid?
Use geohex_grid when you need the uniform shape and size of hexagons for more accurate visual analysis. Geohash cells can vary in shape, while H3 hexagons provide consistency, which is better for calculating densities or distances across the grid.
Does the JWT client authentication replace API keys?
Not necessarily. It provides another, often more standards-based, option for service-to-service authentication. API keys are still a valid and simple solution. JWT authentication is useful if you are already using an Identity Provider that can issue JWTs to your applications.
How does the remote cluster migration assistant work?
It's an internal tool that helps automate the process of switching from a legacy remote cluster configuration to the new remote cluster model based on the transport profile. It handles the reconfiguration steps to minimize downtime and configuration errors during the transition.