version: 9.2.0
released: true
generated: 2025-10-23T12:04:27.512550340Z
changelogs:
  - pr: 74274
    summary: Introduce new rescorer based on script
    area: Search
    type: feature
    issues:
      - 52338
  - pr: 92568
    summary: Support nested fields for term vectors API when using artificial documents
    area: Search
    type: enhancement
    issues:
      - 91902
  - pr: 105718
    summary: Add option for Append Processor to skip/allow empty values
    area: Ingest Node
    type: enhancement
    issues:
      - 104813
  - pr: 113757
    summary: Added AI21 Completion and Chat Completion support to the Inference Plugin
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 113949
    summary: Support kNN filter on nested metadata
    area: Vector Search
    type: enhancement
    issues:
      - 128803
      - 106994
  - pr: 125552
    summary: Expose S3 connection max idle time as a setting
    area: Snapshot/Restore
    type: enhancement
    issues: []
  - pr: 125921
    summary: Allow direct IO for BBQ rescoring
    area: Vector Search
    type: feature
    issues: []
    highlight:
      notable: false
      title: Allow direct IO for BBQ rescoring
      body: |-
        BBQ rescoring performance can be drastically affected by the amount of available
        off-heap RAM for use by the system page cache. When there is not enough off-heap RAM
        to fit all the vector data in memory, BBQ search latencies can be affected by as much as 5000x.
        Specifying the `vector.rescoring.directio=true` Java option on all vector search
        nodes modifies rescoring to use direct IO, which eliminates these very high latencies
        from searches in low-memory scenarios, at a cost of a reduction
        in vector search performance for BBQ indices when the vectors do all fit in memory.

        This option is released in 9.1 as a tech preview whilst we analyse its effect
        for a variety of use cases.
      pr: 125921
  - pr: 127223
    summary: Wrap ES KNN queries with PatienceKNN query
    area: Vector Search
    type: feature
    issues: []
  - pr: 128627
    summary: Fail `profile` on text response formats
    area: ES|QL
    type: feature
    issues: []
  - pr: 128639
    summary: Substitue `date_trunc` with `round_to` when the pre-calculated rounding points are available
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 128866
    summary: Add `age_in_millis` to ILM Explain Response
    area: ILM+SLM
    type: enhancement
    issues:
      - 103659
  - pr: 128917
    summary: Adopt a "LogicalPlan" approach to running multiple sub-queries (with INLINESTATS so far)
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 129003
    summary: Allow timeout during trained model download process
    area: Machine Learning
    type: bug
    issues: []
  - pr: 129013
    summary: Add remote index support to LOOKUP JOIN
    area: ES|QL
    type: feature
    issues: []
    highlight:
      notable: false
      title: Add remote index support to LOOKUP JOIN
      body: |-
        Queries containing LOOKUP JOIN now can be preformed on cross-cluster indices, for example:

        ```yaml
        FROM logs-*, remote:logs-* | LOOKUP JOIN clients on ip | SORT timestamp | LIMIT 100
        ```

      pr: 129013
  - pr: 129090
    summary: Enable force inference endpoint deleting for invalid models and after stopping model deployment fails
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 129146
    summary: "[ML] Add IBM watsonx Completion and Chat Completion support to the Inference Plugin"
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 129369
    summary: Support semantic reranking using contextual snippets instead of entire field text
    area: Relevance
    type: enhancement
    issues: []
  - pr: 129581
    summary: "Support geohash, geotile and geohex grid types"
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 129662
    summary: "[Security] Add entity store and asset criticality index privileges to built in Editor, Viewer and Kibana System roles"
    area: Authorization
    type: enhancement
    issues: []
  - pr: 129693
    summary: Add top level normalizer for linear retriever
    area: Search
    type: enhancement
    issues: []
  - pr: 129848
    summary: "[ML] Add Azure AI Rerank support to the Inference Plugin"
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 129872
    summary: Run `TransportClusterStateAction` on local node
    area: Distributed
    type: enhancement
    issues: []
  - pr: 129929
    summary: Add support for RLIKE (LIST) with pushdown
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 129945
    summary: "Enhancement: ILM sets `indexing_complete` to true from `ReadOnly` action"
    area: ILM+SLM
    type: enhancement
    issues: []
  - pr: 129967
    summary: Support returning default `index_options` for `semantic_text` fields when `include_defaults` is true
    area: Search
    type: bug
    issues: []
  - pr: 129990
    summary: Make forecast write load accurate when shard numbers change
    area: Allocation
    type: bug
    issues: []
  - pr: 130027
    summary: "Fix: prevent duplication of \"invalid index name\" string in the final exception error message"
    area: ES|QL
    type: bug
    issues: []
  - pr: 130092
    summary: Added Llama provider support to the Inference Plugin
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 130140
    summary: Correct slow log user for RCS 2.0
    area: Authentication
    type: enhancement
    issues: []
  - pr: 130251
    summary: Speed up (filtered) KNN queries for flat vector fields
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 130325
    summary: "Simulate API: Return 400 on invalid processor(s)"
    area: Ingest Node
    type: bug
    issues:
      - 120731
    breaking:
      area: Ingest
      title: Return 400 on invalid processor(s) in Simulate API
      details: "In earlier versions of {es}, the Simulate API would return a 500 error when encountering invalid processors. Now, it returns a 400 Bad Request error instead."
      impact: Callers should expect a 400 Bad Request response when the Simulate API encounters invalid processors. This change improves error handling and provides clearer feedback on request issues.
      notable: false
      essSettingChange: false
  - pr: 130330
    summary: "TopNOperator, release Row on failure"
    area: ES|QL
    type: bug
    issues:
      - 130215
  - pr: 130382
    summary: Remove vectors from `_source` transparently
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 130409
    summary: Add Dependency Checker for `LogicalLocalPlanOptimizer`
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 130463
    summary: Refresh potential lost connections at query start for `_search`
    area: Search
    type: enhancement
    issues: []
  - pr: 130485
    summary: Add `RerankRequestChunker`
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 130510
    summary: Add fast path for single value in VALUES aggregator
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 130531
    summary: Adding check for `isIndexed` in text fields when generating field exists queries to avoid ISE when field is stored but not indexed or with `doc_values`
    area: Analysis
    type: bug
    issues: []
  - pr: 130544
    summary: Sync Inference with Trained Model stats
    area: Machine Learning
    type: bug
    issues:
      - 130339
  - pr: 130593
    summary: Add new `CachePopulationReason`
    area: Store
    type: enhancement
    issues: []
  - pr: 130594
    summary: Add audit logging for stream content
    area: Network
    type: enhancement
    issues: []
  - pr: 130635
    summary: "Add low-level optimized Neon, AVX2, and AVX 512 float32 vector operations"
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 130658
    summary: Add support for weighted RRF in retrievers
    area: Relevance
    type: enhancement
    issues: []
  - pr: 130824
    summary: Add read permissions for osquery manager result indices
    area: Security
    type: enhancement
    issues: []
  - pr: 130834
    summary: Ensure vectors are always included in reindex actions
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 130847
    summary: "Pipelines: Add `created_date` and `modified_date`"
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 130855
    summary: Add checks that optimizers do not modify the layout
    area: ES|QL
    type: enhancement
    issues:
      - 125576
  - pr: 130909
    summary: Allow adjustment of transport TLS handshake timeout
    area: Network
    type: enhancement
    issues: []
  - pr: 130939
    summary: Expose HTTP connection metrics to telemetry
    area: Network
    type: enhancement
    issues: []
  - pr: 130940
    summary: Block trained model updates from inference
    area: Machine Learning
    type: enhancement
    issues:
      - 129999
  - pr: 130944
    summary: Remove unnecessary calls to Fold
    area: ES|QL
    type: enhancement
    issues:
      - 119756
  - pr: 130947
    summary: "[main]Prepare Index Like fix for backport to 9.1 and 8.19"
    area: ES|QL
    type: bug
    issues: []
  - pr: 131027
    summary: Handle structured log messages
    area: Ingest Node
    type: feature
    issues:
      - 130333
  - pr: 131056
    summary: Add existing shards allocator settings to failure store allowed list
    area: Data streams
    type: bug
    issues: []
  - pr: 131058
    summary: Adds sparse vector index options settings to semantic_text field
    area: Search
    type: enhancement
    issues: []
  - pr: 131061
    summary: Speed up reading multivalued keywords
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131173
    summary: Add attribute count to `SamlAttribute` `toString`
    area: Authentication
    type: enhancement
    issues: []
  - pr: 131200
    summary: Improve lost-increment message in repo analysis
    area: Snapshot/Restore
    type: enhancement
    issues: []
  - pr: 131204
    summary: Allow pruning columns added by `InlineJoin`
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131261
    summary: Enable Failure Store for new logs-*-* data streams
    area: Data streams
    type: feature
    issues:
      - 131105
    highlight:
      notable: true
      title: Enable Failure Store for new logs data streams
      body: "The [Failure Store](docs-content://manage-data/data-store/data-streams/failure-store.md) is now enabled by default for new logs data streams matching the pattern `logs-*-*`. This means that such data streams will now store invalid documents in a\ndedicated failure index instead of rejecting them, allowing better visibility and control over data quality issues without loosing data. This can be [enabled manually](docs-content://manage-data/data-store/data-streams/failure-store.md#set-up-failure-store-existing) for existing data streams. \nNote: With the failure store enabled, the http response code clients receive when indexing invalid documents will change from `400 Bad Request` to `201 Created`, with an additional response attribute `\"failure_store\" : \"used\"`. "
      pr: 131261
  - pr: 131296
    summary: Enable failure store for newly created APM datastreams
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 131317
    summary: Don't enable norms for fields of type text when the index mode is LogsDB or TSDB
    area: Mapping
    type: breaking
    issues: []
    breaking:
      area: Mapping
      title: Don't enable norms for fields of type text when the index mode is LogsDB or TSDB
      details: "This changes the default behavior for norms on `text` fields in logsdb and tsdb indices. Prior to this change, norms were enabled by default, with the option to disable them via manual configurations. After this change, norms will be disabled by default. Note, because we dont support enabling norms from a disabled state, users will not be able to enable norms on `text` fields in logsdb and tsdb indices."
      impact: Text fields will no longer be normalized by default in LogsDB and TSDB indicies.
      notable: false
      essSettingChange: false
  - pr: 131341
    summary: Consider min/max from predicates when transform date_trunc/bucket to `round_to`
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131390
    summary: Add optimized path for intermediate values aggregator
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131395
    summary: Enable failure store for newly created OTel data streams
    area: Data streams
    type: enhancement
    issues: []
  - pr: 131442
    summary: Track inference deployments
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 131485
    summary: Run single phase aggregation when possible
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131510
    summary: Upgrade apm-agent to 1.55.0
    area: Infra/Metrics
    type: upgrade
    issues: []
  - pr: 131517
    summary: Refresh potential lost connections at query start for field caps
    area: Search
    type: enhancement
    issues: []
  - pr: 131531
    summary: Integrate LIKE/RLIKE LIST with `ReplaceStringCasingWithInsensitiveRegexMatch` rule
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131536
    summary: "Component Templates: Add created and modified date"
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 131551
    summary: Added support to configure query timeout for inference
    area: Inference
    type: enhancement
    issues: []
  - pr: 131559
    summary: Add support for LOOKUP JOIN on multiple fields
    area: ES|QL
    type: enhancement
    issues: []
    highlight:
      notable: true
      title: Add support for Lookup Join on Multiple Fields
      body: |-
        Add support for Lookup Join on Multiple Fields e.g. FROM index1
        | LOOKUP JOIN lookup_index on field1, field2
      pr: 131559
  - pr: 131581
    summary: Support Fields API in conditional ingest processors
    area: Infra/Core
    type: enhancement
    issues: []
  - pr: 131599
    summary: Speed up `OptimizedScalarQuantizer`
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 131641
    summary: Add exception for perform embedding inference requests with query provided
    area: Machine Learning
    type: bug
    issues: []
  - pr: 131694
    summary: Accept unsigned longs on MAX and MIN aggregations
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131706
    summary: Add extension points to remediate index metadata in during snapshot restore
    area: Snapshot/Restore
    type: enhancement
    issues: []
  - pr: 131723
    summary: Tests for FORK's evaluation of field names used in `field_caps` resolve calls
    area: Search
    type: bug
    issues: []
  - pr: 131733
    summary: Replace `RoundTo` linear search evaluator with manual evaluators
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131775
    summary: Replace "representable" type error messages
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 131907
    summary: Enable `exclude_source_vectors` by default for new indices
    area: Vector Search
    type: breaking
    issues: []
    breaking:
      area: Search
      title: Enable `exclude_source_vectors` by default for new indices
      details: |-
        The `exclude_source_vectors` setting is now enabled by default for newly created indices.
        This means that vector fields (e.g., `dense_vector`) are no longer stored in the `_source` field
        by default, although they remain fully accessible through search and retrieval operations.

        Instead of being persisted in `_source`, vectors are now rehydrated on demand from the underlying
        index structures when needed. This reduces index size and improves performance for typical vector
        search workloads where the original vector values do not need to be part of the `_source`.

        If your use case requires vector fields to be stored in `_source`, you can disable this behavior by
        setting `exclude_source_vectors: false` at index creation time.
      impact: |-
        Vector fields will no longer be stored in `_source` by default for new indices. Applications or tools
        that expect to see vector fields in `_source` (for raw document inspection)
        may need to be updated or configured to explicitly retain vectors using `exclude_source_vectors: false`.

        Retrieval of vector fields via search or the `_source` API remains fully supported.
      notable: true
      essSettingChange: false
  - pr: 131937
    summary: Fix race condition in `RemoteClusterService.collectNodes()`
    area: Distributed
    type: bug
    issues: []
  - pr: 131940
    summary: Allow remote enrich after LOOKUP JOIN
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132003
    summary: Add `copy_from` option to the Append processor
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 132011
    summary: Restrict Indexing To Child Streams When Streams Is Enabled
    area: Data streams
    type: enhancement
    issues: []
  - pr: 132048
    summary: Fix stuck in STOPPING by retrying the startup task indefinitely until it succeeds
    area: Transform
    type: bug
    issues:
      - 128221
  - pr: 132064
    summary: Only Allow Enabling Streams If No Conflicting Indices Exist
    area: Data streams
    type: enhancement
    issues: []
  - pr: 132083
    summary: "Index template: Add created_date and modified_date"
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 132131
    summary: Updating `TransportSimulateIndexTemplateAction.resolveTemplate()` to account for data stream overrides
    area: Indices APIs
    type: bug
    issues:
      - 131425
  - pr: 132138
    summary: Fix lookup index resolution when field-caps returns empty mapping
    area: ES|QL
    type: bug
    issues:
      - 132105
  - pr: 132143
    summary: Consider min/max from predicates when transform date_trunc/bucket to `round_to` option 2
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132210
    summary: Adding a `merge_type` parameter to the ingest simulate API
    area: Ingest Node
    type: enhancement
    issues:
      - 131608
  - pr: 132243
    summary: Fix `NullPointerException` in transport trace logger
    area: Network
    type: bug
    issues: []
  - pr: 132321
    summary: "Avoid internal server error when suggester requires unigrams but no unigrams are provided, return bad request instead"
    area: Analysis
    type: bug
    issues:
      - 131928
  - pr: 132362
    summary: Inference API disable partial search results
    area: Machine Learning
    type: bug
    issues: []
  - pr: 132408
    summary: Correct exception for missing nested path
    area: Search
    type: bug
    issues: []
  - pr: 132410
    summary: Add support for retrieving semantic_text's indexed chunks via fields API
    area: Vector Search
    type: feature
    issues: []
  - pr: 132414
    summary: Adjust date docvalue formatting to return 4xx instead of 5xx
    area: Search
    type: bug
    issues: []
  - pr: 132456
    summary: Some optimizations for constant blocks
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132497
    summary: Add cache miss and read metrics
    area: Searchable Snapshots
    type: enhancement
    issues: []
  - pr: 132506
    summary: Add .integration_knowledge system index for usage by AI assistants
    area: Infra/Core
    type: feature
    issues: []
  - pr: 132511
    summary: Handle special regex cases for version fields
    area: Search
    type: bug
    issues: []
  - pr: 132512
    summary: Rewrite `RoundTo` to `QueryAndTags`
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132536
    summary: Avoid stack overflow in `IndicesClusterStateService` `applyClusterState`
    area: Cluster Coordination
    type: bug
    issues: []
  - pr: 132547
    summary: Add epoch blob-cache metric
    area: Searchable Snapshots
    type: enhancement
    issues: []
  - pr: 132548
    summary: Have top level knn searches tracked in query stats
    area: Vector Search
    type: bug
    issues: []
  - pr: 132622
    summary: Push down compute engine value loading of long based singleton numeric doc value to the es819 tsdb doc values codec.
    area: Codec
    type: enhancement
    issues: []
  - pr: 132651
    summary: Use java8 variant of apm-agent
    area: Infra/Core
    type: upgrade
    issues: []
  - pr: 132675
    summary: Add second max queue latency stat to `ClusterInfo`
    area: Allocation
    type: enhancement
    issues: []
  - pr: 132680
    summary: Add support for per-field weights in simplified RRF retriever syntax
    area: Search
    type: enhancement
    issues: []
  - pr: 132689
    summary: Add support for dimensions in google vertex ai request
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 132738
    summary: Fix `AsyncOperator` status values and add emitted rows
    area: ES|QL
    type: bug
    issues: []
  - pr: 132744
    summary: Fix async operator warnings not always sent when blocking
    area: ES|QL
    type: bug
    issues:
      - 130642
      - 132554
      - 132778
      - 130296
      - 132555
      - 131563
      - 131148
      - 132604
      - 128030
  - pr: 132765
    summary: Implement `v_magnitude` function
    area: ES|QL
    type: feature
    issues:
      - 132768
  - pr: 132774
    summary: Improve cpu utilization with dynamic slice size in doc partitioning
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132833
    summary: Adding simulate ingest effective mapping
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 132845
    summary: Expose existing DLS cache x-pack usage statistics
    area: Authorization
    type: enhancement
    issues: []
  - pr: 132858
    summary: Add index mode to resolve index response
    area: Indices APIs
    type: feature
    issues: []
  - pr: 132934
    summary: Support filters on inlinestats
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132950
    summary: Speed up loading keyword fields with index sorts
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 132959
    summary: Adds the `v_hamming` function for calculating the Hamming distance between two dense vectors
    area: ES|QL
    type: feature
    issues:
      - 132056
  - pr: 132967
    summary: ES-11331 streams params restriction
    area: Data streams
    type: enhancement
    issues: []
  - pr: 133016
    summary: Adding Contains ESQL String function
    area: ES|QL
    type: feature
    issues: []
  - pr: 133018
    summary: Add ordinal range encode for tsid
    area: TSDB
    type: enhancement
    issues: []
  - pr: 133030
    summary: Safely prevent overwriting objects in S3 repositories
    area: Snapshot/Restore
    type: enhancement
    highlight:
      notable: true
      title: Safely prevent overwriting objects in S3 repositories
      body: |-
        Earlier versions of Elasticsearch had a small risk of inadvertently overwriting an object in an [AWS S3 snapshot
        repository](docs-content://deploy-manage/tools/snapshot-and-restore/s3-repository.md) which, if it happened, might corrupt the
        repository contents. From version 9.2.0 onwards, Elasticsearch uses S3's [conditional write
        feature](https://docs.aws.amazon.com/AmazonS3/latest/userguide/conditional-writes.html) to prevent this corruption.

        If you are using a snapshot repository with type `s3` backed by some storage which is not AWS S3, but which is fully S3-compatible,
        then you will also get the benefits of this corruption protection by upgrading to 9.2.0 or later. Refer to [S3-compatible
        services](docs-content://deploy-manage/tools/snapshot-and-restore/s3-repository.md#repository-s3-compatible-services) for more
        information about using Elasticsearch with S3-compatible storage services.
      pr: 133030
    issues:
      - 128565
  - pr: 133064
    summary: Mark LOOKUP JOIN as `ExecutesOn.Any` by default
    area: ES|QL
    type: bug
    issues: []
  - pr: 133074
    summary: Adds transport-only flag to always include indices in the field caps transport response
    area: Mapping
    type: enhancement
    issues: []
  - pr: 133087
    summary: Esql skip null metrics
    area: ES|QL
    type: enhancement
    issues:
      - 129524
  - pr: 133111
    summary: Return 429 instead of 500 for timeout handlers
    area: Network
    type: enhancement
    issues: []
  - pr: 133154
    summary: Allow configuring SAML private attributes
    area: Authentication
    type: enhancement
    issues: []
  - pr: 133166
    summary: Improve Expanding Lookup Join performance by pushing a filter to the right side of the lookup join
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 133193
    summary: Fix offset handling in Murmur3Hasher
    area: Infra/Core
    type: bug
    issues: []
  - pr: 133245
    summary: Add query heads priority to `SliceQueue`
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 133314
    summary: Add hits and misses timing stats to DLS cache
    area: Authorization
    type: enhancement
    issues: []
  - pr: 133357
    summary: Switch to Sending a Bad Request User When Function Score Query Generates Negative Scores
    area: Search
    type: bug
    issues:
      - 133358
  - pr: 133360
    summary: Add recover_failure_document processor to remediate failurestore docs
    area: Ingest Node
    type: feature
    issues: []
  - pr: 133365
    summary: Skip iterating DISI when reading metric values
    area: Codec
    type: enhancement
    issues: []
  - pr: 133369
    summary: Enable `date` `date_nanos` implicit casting
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 133397
    summary: Push down loading of singleton dense double based field types to the …
    area: Codec
    type: enhancement
    issues: []
  - pr: 133403
    summary: Fix service destination template file name
    area: Data streams
    type: bug
    issues: []
  - pr: 133546
    summary: "Support geohash, geotile and geohex grid types in ST_INTERSECTS and ST_DISJOINT"
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 133576
    summary: Text similarity reranker chunks and scores snippets
    area: Relevance
    type: enhancement
    issues: []
  - pr: 133599
    summary: Support Gemini thinking budget in inference API
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 133601
    summary: Fix bug in topn
    area: ES|QL
    type: bug
    issues:
      - 133600
      - 133574
      - 133607
  - pr: 133616
    summary: Add mode filter to _resolve/index
    area: Indices APIs
    type: enhancement
    issues: []
  - pr: 133636
    summary: Esql `mv_contains` function
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 133659
    summary: Fixed GeneralScriptException to return 400 http status code
    area: Infra/Scripting
    type: bug
    issues: []
  - pr: 133675
    summary: Support using the semantic query across multiple inference IDs
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 133683
    summary: Avoid running asynchronous ILM actions while ILM is stopped
    area: ILM+SLM
    type: bug
    issues:
      - 99859
      - 81234
      - 85097
  - pr: 133718
    summary: Remove upper limit for chunking settings
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 133720
    summary: Support querying multiple indices with the simplified linear retriever
    area: Relevance
    type: enhancement
    issues: []
  - pr: 133722
    summary: Fix disk usage estimation for SORTED_SET doc values
    area: Codec
    type: bug
    issues: []
  - pr: 133745
    summary: Supporting more timestamp formats in `_text_structure/find_structure`
    area: Machine Learning
    type: feature
    issues: []
  - pr: 133860
    summary: Cache Inference Endpoints
    area: Machine Learning
    type: enhancement
    issues:
      - 133135
  - pr: 133861
    summary: Implementing latency improvements for EIS integration
    area: Machine Learning
    type: bug
    issues: []
  - pr: 133897
    summary: "Runtime fields: pass down runtime field name as source filter when source mode is synthetic"
    area: Mapping
    type: enhancement
    issues: []
  - pr: 133916
    summary: Fix model assignment error handling and assignment explanation generation
    area: Machine Learning
    type: bug
    issues: []
  - pr: 133930
    summary: Improve memory estimation methods accuracy in `TrainedModelAssignmentRebalancer` and related classes
    area: Machine Learning
    type: bug
    issues: []
  - pr: 133952
    summary: Fix warning when creating an OTel data stream
    area: TSDB
    type: bug
    issues:
      - 132918
  - pr: 133954
    summary: "ILM: Force merge on zero-replica cloned index before snapshotting for searchable snapshots"
    area: ILM+SLM
    type: enhancement
    issues:
      - 75478
  - pr: 133968
    summary: Extend kibana-system permissions to manage security entities
    area: Infra/Core
    type: enhancement
    issues: []
  - pr: 133980
    summary: Upgrade elasticsearch to lucene 10.3.0
    area: Search
    type: upgrade
    issues: []
    highlight:
      notable: false
      title: New lucene 10.3.0 release
      body: |-
        - Improved performance for lexical, vector and primary-key searches
        - Use optimistic-with-checking KNN Query execution strategy in place of cross-thread global queue min-score checking. Improves performance and consistency.
        - Bulk scoring added for floating point vectors in HNSW. Improving query latency and indexing throughput
        - Multiple improvements to HNSW graph traversal and storage
      pr: 133980
  - pr: 134029
    summary: Add SET instruction
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 134033
    summary: Fix FORK with union-types
    area: ES|QL
    type: bug
    issues:
      - 133973
  - pr: 134080
    summary: Added Google Model Garden Anthropic Completion and Chat Completion support to the Inference Plugin
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 134098
    summary: LOOKUP JOIN with expressions
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 134117
    summary: Improve block loader for source only runtime fields of type long
    area: Mapping
    type: enhancement
    issues: []
  - pr: 134137
    summary: Add signing configuration for cross cluster api keys
    area: Security
    type: enhancement
    issues: []
  - pr: 134152
    summary: Correctly update SLM stats with master shutdown
    area: ILM+SLM
    type: bug
    issues: []
  - pr: 134178
    summary: Return 429 status when `RequestExecutorService` queue full
    area: Machine Learning
    type: bug
    issues: []
  - pr: 134198
    summary: Improve `ShardLockObtainFailedException` message
    area: Store
    type: enhancement
    issues: []
  - pr: 134214
    summary: "[Downsampling++] Add time series telemetry in xpack usage"
    area: Downsampling
    type: enhancement
    issues:
      - 133953
  - pr: 134232
    summary: Add relevant attributes to search took time APM metrics
    area: Search
    type: enhancement
    issues: []
  - pr: 134309
    summary: Telemetry with inlinestats
    area: ES|QL
    type: bug
    issues: []
  - pr: 134317
    summary: ES|QL - Allow multivalued query parameters
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 134323
    summary: Make INLINESTATS (and subplans) work with CCS
    area: ES|QL
    type: enhancement
    issues:
      - 124748
  - pr: 134349
    summary: Add `LoadedSecureSettings` for keeping temporary secure settings loaded
    area: Security
    type: enhancement
    issues: []
  - pr: 134365
    summary: Remove `PeerFinder` request timeout
    area: Cluster Coordination
    type: deprecation
    issues: []
    deprecation:
      area: Cluster and node setting
      title: Remove `PeerFinder` request timeout
      details: "There is no need to time out requests sent by the `PeerFinder` during discovery and cluster formation, and this timeout may sometimes cause spurious failures. With this change the `PeerFinder` requests will wait indefinitely for responses. The `discovery.request_peers_timeout` setting no longer has any effect."
      impact: Discontinue use of the `discovery.request_peers_timeout` setting.
      notable: false
      essSettingChange: false
  - pr: 134374
    summary: Add file extension metadata to cache miss counter from `SharedBlobCacheService`
    area: Search
    type: enhancement
    issues: []
  - pr: 134415
    summary: Differentiate between initial and reconnect RCS connections
    area: Network
    type: enhancement
    issues: []
  - pr: 134446
    summary: Add support for `include_execution_metadata` parameter
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 134457
    summary: Improve scalability of get-license action
    area: License
    type: enhancement
    issues: []
  - pr: 134458
    summary: Enable caching of all filters in `knn` queries
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 134475
    summary: ES|QL Absent and `AbsentOverTime` functions
    area: ES|QL
    type: enhancement
    issues:
      - 131069
  - pr: 134497
    summary: Limit when we push topn to lucene
    area: ES|QL
    type: bug
    issues: []
  - pr: 134504
    summary: Adding headers support for OpenAI chat completion
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 134524
    summary: Add support for flexible access pattern to `NormalizeForStreamProcessor`
    area: Ingest Node
    type: bug
    issues: []
  - pr: 134585
    summary: Add classes to represent raw docs sampling configs
    area: Ingest Node
    type: feature
    issues: []
  - pr: 134609
    summary: Add 'SslProfileExtension' SPI interface
    area: TLS
    type: enhancement
    issues: []
  - pr: 134629
    summary: Improve block loader for source only runtime fields of type double
    area: Mapping
    type: enhancement
    issues: []
  - pr: 134717
    summary: Allow including semantic field embeddings in `_source`
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 134798
    summary: Add relevant attributes to shard search latency APM metrics
    area: Search
    type: enhancement
    issues: []
  - pr: 134822
    summary: Support querying multiple indices with the simplified RRF retriever
    area: Relevance
    type: enhancement
    issues: []
  - pr: 134835
    summary: Add new `/_security/stats` endpoint
    area: Authorization
    type: enhancement
    issues: []
  - pr: 134851
    summary: Remove ingest conditionals `_type` deprecation warning
    area: Ingest Node
    type: enhancement
    issues: []
  - pr: 134879
    summary: Throw 4xx instead of 5xx for ESQL malformed query params
    area: ES|QL
    type: bug
    issues:
      - 134618
  - pr: 134893
    summary: Add trust configuration for cross cluster api keys
    area: Security
    type: enhancement
    issues: []
  - pr: 134894
    summary: "[Cyera] Add `manage`, `create_index`, `read`, `index`, `write`, `delete`, permission for third party agent indices `kibana_system`"
    area: Authorization
    type: enhancement
    issues:
      - 134183
  - pr: 134933
    summary: Add ContextualAI Rerank Service Implementation to the Inference API
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 134936
    summary: Fixing conditional processor mutability bugs
    area: Ingest Node
    type: bug
    issues: []
  - pr: 134941
    summary: "DLM: Better `max_age` rollover for tiny retentions"
    area: Data streams
    type: enhancement
    issues:
      - 130960
  - pr: 134942
    summary: Add telemetry support for Lookup Join On Expression
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 134952
    summary: Add support for expressions with LOOKUP JOIN in tech preview
    area: ES|QL
    type: enhancement
    issues: []
    highlight:
      notable: true
      title: Add support for expressions with LOOKUP JOIN in tech preview
      body: |-
        Enable Lookup Join on Expression Tech Preview
        FROM index1 | LOOKUP JOIN lookup_index on left_field1 > right_field1 AND left_field2 <= right_field2
      pr: 134952
  - pr: 134960
    summary: Adding custom headers support openai text embeddings
    area: Machine Learning
    type: enhancement
    issues: []
  - pr: 135011
    summary: Handle right hand side of Inline Stats coming optimized with `LocalRelation` shortcut
    area: ES|QL
    type: bug
    issues: []
  - pr: 135026
    summary: Improve block loader for source only runtime fields of type keyword
    area: Mapping
    type: enhancement
    issues: []
  - pr: 135036
    summary: Performance improvements for Lookup Join on Expression
    area: ES|QL
    type: enhancement
    issues: []
  - pr: 135039
    summary: Use optimized field visitor for ignored source queries
    area: Mapping
    type: enhancement
    issues: []
  - pr: 135053
    summary: Make `SecureString` comparisons constant time
    area: Infra/Core
    type: enhancement
    issues: []
  - pr: 135087
    summary: "Optimize `BytesArray::indexOf,` which is used heavily in ndjson parsing"
    area: Performance
    type: feature
    issues: []
  - pr: 135097
    summary: Fixes `countDistinctWithConditions` in csv-spec tests
    area: ES|QL
    type: bug
    issues:
      - 134380
  - pr: 135198
    summary: Enable `chunk_rescorer` in `text_similarity_reranker`
    area: Relevance
    type: enhancement
    issues: []
  - pr: 135204
    summary: Make `_tsid` available in metadata
    area: ES|QL
    type: enhancement
    issues:
      - 133205
  - pr: 135244
    summary: Add reload listener to `SslProfile`
    area: TLS
    type: enhancement
    issues: []
  - pr: 135247
    summary: Fix alias id when drop all aggregates
    area: ES|QL
    type: bug
    issues: []
  - pr: 135262
    summary: Add usage stats for `semantic_text` fields
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 135263
    summary: Optimize `dotCount` in expanding dot parser
    area: Mapping
    type: enhancement
    issues: []
  - pr: 135271
    summary: Add DLS stats to `_security/stats`
    area: Authorization
    type: enhancement
    issues: []
  - pr: 135295
    summary: Replace any Attribute type when pushing down past Project
    area: ES|QL
    type: bug
    issues:
      - 134407
  - pr: 135299
    summary: Release DiskBBQ(`bbq_disk`) index type for `dense_vector` fields
    area: Vector Search
    type: feature
    issues: []
    highlight:
      notable: true
      title: Release DiskBBQ(`bbq_disk`) index type for `dense_vector` fields
      body: |-
        This provides a new index type called DiskBBQ (`bbq_disk`).
        DiskBBQ is a cluster based format that provides:
          - faster and cheaper indexing than HNSW
          - Better behavior in lower memory environments (degrades linearly, not exponentially)
          - Is near HNSW for QPS when the index is in memory

        Current restrictions:
          - only floating point values are allowed currently
          - quantization is only to a single bit, so not recommended for low dimensionality vectors
          - all other restrictions that exist for `dense_vector` fields still apply

        To utilize the format, its just like any other:

        ```yaml
        PUT vectors
         {
           "mappings": {
             "properties": {
               "vector": {"type": "dense_vector", "index_options": {"type": "disk_bbq"}
             }
           }
         }
        ```

         Querying is just like any other field.

        ```yaml
        POST vectors/_search{
          "query": {
            "knn": {
              "field": "vector",
              "query_vector": <vector>,
              "k": 3
            }
          }
        }
        ```

        `num_candidates` can be used for tuning approximate nature of the search.
        Or, more granular control can be provided by setting `visit_percentage` directly.
      pr: 135299
  - pr: 135306
    summary: Add support for extended search usage telemetry
    area: Relevance
    type: enhancement
    issues: []
  - pr: 135309
    summary: Enable semantic search CCS when ccs_minimize_roundtrips=true
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 135337
    summary: Do not pass `ProjectMetadata` to lazy index permissions builder
    area: Security
    type: enhancement
    issues: []
  - pr: 135342
    summary: Add 'profile' support for knn query on HNSW with early termination
    area: Vector Search
    type: enhancement
    issues: []
  - pr: 135370
    summary: Add new `pattern_text` field mapper in tech preview
    area: Mapping
    type: feature
    issues: []
  - pr: 135373
    summary: Improve block loader for source only runtime date fields
    area: Mapping
    type: enhancement
    issues: []
  - pr: 135393
    summary: Improve block loader for source only runtime IP fields
    area: Mapping
    type: enhancement
    issues: []
  - pr: 135401
    summary: Adds an OTLP metrics endpoint (`_otlp/v1/metrics`) as tech preview
    area: TSDB
    type: enhancement
    issues: []
  - pr: 135402
    summary: "Improve TSDB ingestion by hashing dimensions only once, using a new auto-populeted `index.dimensions` private index setting"
    area: TSDB
    type: enhancement
    issues: []
  - pr: 135403
    summary: Take INLINE STATS out of snapshot
    area: ES|QL
    type: enhancement
    issues: []
    highlight:
      notable: true
      title: Enable INLINE STATS in non-snapshot builds
      body: |-
        This effectively releases INLINE STATS into tech preview.
        - Enable the lexing/grammar for INLINE STATS in non-snapshot builds.
        - Enable more tests with FORK and INLINE STATS
      pr: 135403
  - pr: 135431
    summary: "[Downsampling++] Allow merging of passthrough mappers with object mappers under certain conditions"
    area: Mapping
    type: bug
    issues: []
  - pr: 135446
    summary: Fix projection generation when pruning left join
    area: ES|QL
    type: bug
    issues: []
  - pr: 135484
    summary: Add origin to client in SLM task
    area: ILM+SLM
    type: bug
    issues: []
  - pr: 135524
    summary: Add time range bucketing attribute to APM shard search latency metrics
    area: Search
    type: enhancement
    issues: []
  - pr: 135545
    summary: Add GPUPlugin for indexing vectors on GPU
    area: Vector Search
    type: feature
    issues: []
  - pr: 135547
    summary: Fix union types lost attributes in `StubRelation` for inlinestats
    area: ES|QL
    type: bug
    issues: []
  - pr: 135603
    summary: Make FUSE available in release builds
    area: ES|QL
    type: feature
    issues: []
  - pr: 135604
    summary: ES|QL - add `dense_vector` field type
    area: ES|QL
    type: feature
    issues: []
  - pr: 135635
    summary: Add executor name attribute to cache miss metrics
    area: Search
    type: enhancement
    issues: []
  - pr: 135652
    summary: DFS search phase per shard duration APM metric
    area: Search
    type: enhancement
    issues: []
  - pr: 135673
    summary: Add index setting that disables the `index.dimensions` based routing and `_tsid` creation strategy
    area: TSDB
    type: enhancement
    issues: []
  - pr: 135709
    summary: ES|QL - KNN function
    area: ES|QL
    type: feature
    issues: []
  - pr: 135767
    summary: Pass fix size instead of `maxPageSize` to `LuceneTopNOperator` scorer
    area: ES|QL
    type: bug
    issues: []
  - pr: 135838
    summary: Remove rate limit field from services API for EIS
    area: Machine Learning
    type: bug
    issues: []
  - pr: 135897
    summary: Apply source excludes early when retrieving the `_inference_fields`
    area: Search
    type: bug
    issues: []
  - pr: 135966
    summary: Ensure queued `AbstractRunnables` are notified when executor stops
    area: Machine Learning
    type: bug
    issues:
      - 134651
  - pr: 135987
    summary: Avoid rewrite `round_to` with expensive queries
    area: ES|QL
    type: bug
    issues: []
  - pr: 135994
    summary: Cardinality Aggregator Throws `UnsupportedOperationException` When Field Type is Vector
    area: Vector Search
    type: bug
    issues: []
  - pr: 136017
    summary: Cohere service Model Id field is required
    area: Machine Learning
    type: bug
    issues: []
  - pr: 136030
    summary: Update to lucene 10.3.1
    area: Search
    type: feature
    issues: []
  - pr: 136058
    summary: Configurable HTTP read and connect timeouts for url based SAML metadata resolution
    area: Security
    type: bug
    issues: []
  - pr: 136105
    summary: Fix a breaker bug
    area: ES|QL
    type: bug
    issues:
      - 135224
      - 135260
  - pr: 136119
    summary: Fix logsdb settings provider mapping filters
    area: Logs
    type: bug
    issues:
      - 136107
  - pr: 136214
    summary: Using index setting providers for data stream setting validation
    area: Data streams
    type: bug
    issues:
      - 136166
  - pr: 136279
    summary: Initialize `TermsEnum` eagerly
    area: Search
    type: bug
    issues: []
  - pr: 136312
    summary: Fix _inference_fields handling on old indices
    area: Vector Search
    type: bug
    issues:
      - 136130
  - pr: 136315
    summary: Store full path in `_ignored` when ignoring dynamic array field
    area: Mapping
    type: bug
    issues: []
  - pr: 136348
    summary: Add error message when using inline stats on TS
    area: ES|QL
    type: bug
    issues:
      - 136092
  - pr: 136444
    summary: Create new block when filter `OrdinalBytesRefBlock`
    area: ES|QL
    type: bug
    issues:
      - 136423
  - pr: 136492
    summary: Make `ResolveUnionTypes` rule stateless
    area: ES|QL
    type: bug
    issues: []
  - pr: 136501
    summary: "Reindex-from-remote: Validate basic auth params"
    area: Indices APIs
    type: bug
    issues:
      - 135925
  - pr: 136556
    summary: Bugfix 136545
    area: Vector Search
    type: bug
    issues: []
  - pr: 136563
    summary: Bugfix/disable matches highlight knn
    area: Vector Search
    type: bug
    issues: []
  - pr: 136577
    summary: Clean up inference indices on failed endpoint creation
    area: Machine Learning
    type: bug
    issues:
      - 123726
  - pr: 136625
    summary: Optimize Index Permission Automatons for Has Privileges
    area: Security
    type: bug
    issues: []
  - pr: 136649
    summary: Fix append processor `ignore_empty_values` edge case
    area: Ingest Node
    type: bug
    issues: []
  - pr: 136664
    summary: Drop project-id from threadcontext for CCS
    area: Authorization
    type: bug
    issues: []
  - pr: 136684
    summary: Lazy compute and cache `grantsAll` per privilege
    area: Authorization
    type: enhancement
    issues: []
  - pr: 136709
    summary: Fix log formatting in `SnapshotLifecycleTask`
    area: ILM+SLM
    type: bug
    issues: []
  - pr: 136751
    summary: Adjust jinaai rerank response parser to handle document field as string or object
    area: Machine Learning
    type: bug
    issues: []
  - pr: 136759
    summary: Avoid counting snapshot failures twice in SLM
    area: ILM+SLM
    type: bug
    issues: []
  - pr: 136769
    summary: Release cluster state
    area: Machine Learning
    type: bug
    issues:
      - 123243