🌈 Enhancements

Search personalization on federated search

We now support using the search personalization feature on federated search requests.

Like page/hitPerPage or limit/offset, the personalization option must be specified in the federation attribute to work properly.
Otherwise, an error will be returned reminding you to move the attribute in federation.

By @ManyTheFish in #6414

The new settings indexer is feature complete 🎉

• We now better support the tokenizer-related settings
• We improved the quality of the new settings indexer to enhance the engine's performance when changing the locales, the dictionary, synonyms, stop words, separator, and non-separator tokens.
• This makes the new settings indexer feature-complete, meaning that, unless you set the MEILI_EXPERIMENTAL_NO_EDITION_2024_FOR_SETTINGS environment variable to true, all settings tasks can now be handled by the new settings indexer, bringing a better scaling behavior, much faster cancellation, and a more precise progress view.

By @Kerollmops in #6409

Observability improvements

We expose more Prometheus metrics to improve observability, specifically to show more metrics on document throughput and ease debugging.

By @Kerollmops in #6430

🦋 Fixes

• Putting attributeRank/ wordPosition before words in the rankingRules list will no longer remove hits from the response, by @pjdurden in #6437
• Meilisearch will no longer ignore the searchCutoffMs in some conditions when embedding documents, by @dureuill in #6447
• Meilisearch will no longer fail to proxy a search request with a filter containing a ' during remote federated search or useNetwork: true search requests, by @dureuill in #6445

🔒 Security

• CI: Prevent shell injection in GitHub Actions release workflows by @curquiza in #6420

🔩 Misc. changes

Search implementation refactor

Refactor the search pipeline to mutualize the code.
The new implementation will always perform a federated search under the hood, and then the output will be transformed into the expected route's output.

Noticeable changes from the user perspective:

• Some error messages have been modified
• Small breaking change: a few error codes change, such as MultiSearch<Error> <--> Search<Error>

Other changes

• Make it easier and less error-prone to declare a type that is used as a body parameter on a Meilisearch endpoint, by @dureuill in #6429
• Remove the now unused vectorStoreBackend setting from the settings, by @Kerollmops in #6399
• Replace custom hf-hub git dependency by the official one by @Kerollmops in #6442
• Update Python SDK test CI by @Strift in #6439
• Refactor the code to use the MustStopProcessing type everywhere by @Kerollmops in #6423

New Contributors

• @pjdurden made their first contribution in #6437

Full Changelog: v1.46.1...v1.47.0

Queue documents fetch routes

Add an experimental feature, queueDocumentsFetch, forcing the routes GET indexes/:uid/documents and POST indexes/:uid/documents/fetch to wait in the search queue if there is no available thread to process them.

This release introduces fixes for a regression in v1.45.0, where we were batching deletions by filter with other deletions or additions. It also enables the new settings indexer to support more parameters, making the engine faster to index documents when those settings are specified.

✨ Enhancement

• Support exact and disable on numbers in the new settings indexer by @Kerollmops in #6398
  Introduce support for exact words and disable-on-words parameters in the new settings indexer, making the engine more efficient when changing these settings.

• Support computing prefixes in the new settings indexer by @Kerollmops in #6391
  Support for the prefix search settings in the new settings indexer, so that changing this parameter makes the engine more efficient.

🪲 Bug fixes

• Better limit read bytes when creating the S3 multipart part by @Kerollmops in #6405
  This fixes an issue we had with the multipart part size by ensuring we never construct a part larger than the defined multipart part size. With this fix, we always create a multipart with the provided multipart part size, except for the last part. Thanks, @vidit-virmani, for the help investigating the issue.

• Batch of documentDeletionByFilter with documentAdditionOrUpdate by @Kerollmops and @ManyTheFish in #6415
  Correctly implement the support for auto-batching deletion by filter with document replacement and updates. This fixes a regression introduced in v1.45.0.

• Fix a panic with incomplete filters by @Kerollmops in #6421
  Fixes an internal panic when a filter is incomplete by returning an error instead.

🔒 Security

• Bump tar from 0.4.45 to 0.4.46 by @dependabot[bot] in #6418

🔩 Miscellaneous

• OpenAPI CI: disable rule to avoid crash (workaround) by @curquiza in #6417
• Remove milli benchmarks by @Kerollmops in #6419
• Add precision in the index swap documentation by @ManyTheFish in #6408
• Fix broken links in the documentation by @Kerollmops in #6406

🪲 Bug Fixes

• Fix internal vector stores quantization config desync by @Kerollmops in #6411
  We noticed issues when dumpless upgrading databases containing embedders from versions older than v1.33.1. This version fixes those databases by removing the corrupted embeddings from the impacted indexes.

Full Changelog: v1.45.1...v1.45.2

🪲 Bug Fixes

• Revert autobatch deletions by filter with additions by @Kerollmops in #6412

Full Changelog: v1.45.0...v1.45.1

Meilisearch v1.45.0 mainly improves indexing performance when changing settings and also improves document fetch performance.

✨ Enhancement

• Autobatch deletions by filter with additions by @Kerollmops in #6389
  While Meilisearch tries to improve indexing speed when users add documents and mix those additions with deletions via a filter, it is still an anti-pattern to interleave both. It is recommended to delete documents by ID whenever possible, as the engine is optimized to merge them, thereby drastically speeding up indexation.
• Improve settings indexing performance & visibility
  We are introducing support for more settings in the nez settings indexer. If you find any bugs, please report them on GitHub. You can disable the new settings indexer by setting the environment variable like this: MEILI_EXPERIMENTAL_NO_EDITION_2024_FOR_SETTINGS=true or use the equivalent dedicated CLI parameter.
  • Support non-extracting parameters by @Kerollmops in #6393
    Changing one of the following settings is now directly handled by the new settings indexer: displayed fields, synonyms, the primary key, authorize typos, min word len one and two typos, max values per facet, sort facet values by, pagination max total hits, search cut off, chat, and foreign keys.
  • Support the global facet search by @Kerollmops in #6390
    Meilisearch can use the new settings indexer when the settings change the facet search root boolean parameter.

🪲 Bug Fixes

• Fetching documents no longer blocks the actix worker @dureuill in #6402
• Fix an internal error when changing the binary quantization by @Kerollmops in #6396
  We fixed an issue where users were changing the binary-quantized boolean in embedder configurations. The change corrupted the database, making it impossible to change the quantization in the future. Users had to create a binary-quantized embedder from scratch, or they could never change it again.

🔩 Miscellaneous

• Make timeout test less flaky on Windows by @dureuill in #6388

Full Changelog: v1.44.0...v1.45.0

Meilisearch v1.44.0 adds remote federated facet search, indexing performance improvements, and other improvements and bugfixes. It also contains a couple of breaking changes, detailed below.

Breaking changes

• When using the network experimental feature, with sharding enabled (leader is not null in the network configuration), POST /indexes/{indexUid}/facet-search calls now default to a remote federated facet search, fetching and merging results from all shards in the network.
• The timeout for calling an external REST embedder at search time is now tied to the searchCutOffMs defined in the index, rather than fixed.
  • If you observe missing semantic results after upgrading, or HTTP 500 errors for pure semantic search, please increase the value of the search cutoff in the index settings.
  • By @dureuill in #6377

🌈 Improvements

Remote facet search

Meilisearch now has the ability to search across all shards of a network during facet search via the existing dedicated facet search route.

• If you are using the network experimental feature and have a leader defined for your network, remote calls are now the default for calls to POST /indexes/{indexUid}/facet-search
• The behavior can be controlled explicitly via the new useNetwork parameter of the facet search object.

By @dureuill in #6375

Reduce memory usage of the indexing

Reduces allocated memory when computing prefixes and speeds up some operations to avoid unnecessary deserialization.

If you still see high memory usage while the engine is post-processing, we recommend using the --experimental-reduce-indexing-memory-usage option.

By @Kerollmops in #6334

Improve GeoJSON indexing performance

This PR upgrades the cellulite library to a version that includes a GeoJSON indexing optimization. With this change, GeoJSON indexing avoids reprocessing documents that were already indexed in dense cells, and only handles newly added documents incrementally as they descend through the recursive cell tree.

By @YoEight in #6374

Human-formatted sizes and detailed DB sizes in stats

Adds two new query parameters to GET /indexes/{indexUid}/stats and GET /stats:

• showInternalDatabaseSizes: boolean, optional, defaults to false. When present, the index stat objects in responses of the stat routes now contain an additional internalDatabaseSizes key, whose value is a dictionary of the internal database names and their current size in the index, like in the stats object of a batch.
• sizeFormat: human or raw: optional, defaults to raw. When present and set to human, then all database sizes in responses of the stat routes will be returned as a string containing an appropriate unit (MiB, GiB, etc). When missing or set to raw, then the current behavior of expressing the size in bytes as a number is retained.

Note for integrations: The keys in internalDatabaseSizes are subject to change and should not be exposed as a strongly typed object. This is the same as the existing internalDatabaseSizes in batch stats.

// curl -X GET "http://localhost:7700/indexes/movies/stats?showInternalDatabaseSizes=true"
{
  "numberOfDocuments": 31944,
  "rawDocumentDbSize": 20594688,
  "avgDocumentSize": 636,
  "isIndexing": false,
  "internalDatabaseSizes": {
    "wordPairProximityDocids": 100827136,
    "documents": 20594688,
    "wordPositionDocids": 18694144,
    "wordFidDocids": 10715136,
    "wordPrefixPositionDocids": 10256384,
    "wordDocids": 9453568,
    "wordPrefixFidDocids": 4603904,
    "wordPrefixDocids": 3424256,
    "main": 1425408,
    "externalDocumentsIds": 999424,
    "fieldIdWordCountDocids": 245760,
    "exactWordPrefixDocids": 16384,
    "celluliteMetadata": 16384
  },
  "numberOfEmbeddings": 0,
  "numberOfEmbeddedDocuments": 0,
  "fieldDistribution": {
    "genres": 31944,
    "id": 31944,
    "overview": 31944,
    "poster": 31944,
    "release_date": 31944,
    "title": 31944
  }
}

// curl -X GET "http://localhost:7700/stats?showInternalDatabaseSizes=true&sizeFormat=human"
{
  "databaseSize": "351.38 MiB",
  "usedDatabaseSize": "350.64 MiB",
  "lastUpdate": "2026-04-16T13:07:02.74243Z",
  "noindexes": {
    "comics": {
      "numberOfDocuments": 31944,
      "rawDocumentDbSize": "23.14 MiB",
      "avgDocumentSize": "751 B",
      "isIndexing": false,
      "internalDatabaseSizes": {
        "wordPairProximityDocids": "98.06 MiB",
        "documents": "23.14 MiB",
        "wordPositionDocids": "17.22 MiB",
        "wordFidDocids": "10.36 MiB",
        "wordPrefixPositionDocids": "9.47 MiB",
        "wordDocids": "8.97 MiB",
        "wordPrefixFidDocids": "4.36 MiB",
        "wordPrefixDocids": "3.27 MiB",
        "main": "1.36 MiB",
        "externalDocumentsIds": "944 KiB",
        "fieldIdWordCountDocids": "240 KiB",
        "exactWordPrefixDocids": "16 KiB",
        "celluliteMetadata": "16 KiB"
      },
      "numberOfEmbeddings": 0,
      "numberOfEmbeddedDocuments": 0,
      "fieldDistribution": {
        "genres": 31944,
        "id": 31944,
        "overview": 31944,
        "poster": 31944,
        "release_date": 31944,
        "title": 31944
      }
    },
    "movies": {
      "numberOfDocuments": 31944,
      "rawDocumentDbSize": "19.64 MiB",
      "avgDocumentSize": "636 B",
      "isIndexing": false,
      "internalDatabaseSizes": {
        "wordPairProximityDocids": "96.16 MiB",
        "documents": "19.64 MiB",
        "wordPositionDocids": "17.83 MiB",
        "wordFidDocids": "10.22 MiB",
        "wordPrefixPositionDocids": "9.78 MiB",
        "wordDocids": "9.02 MiB",
        "wordPrefixFidDocids": "4.39 MiB",
        "wordPrefixDocids": "3.27 MiB",
        "main": "1.36 MiB",
        "externalDocumentsIds": "976 KiB",
        "fieldIdWordCountDocids": "240 KiB",
        "exactWordPrefixDocids": "16 KiB",
        "celluliteMetadata": "16 KiB"
      },
      "numberOfEmbeddings": 0,
      "numberOfEmbeddedDocuments": 0,
      "fieldDistribution": {
        "genres": 31944,
        "id": 31944,
        "overview": 31944,
        "poster": 31944,
        "release_date": 31944,
        "title": 31944
      }
    }
  }
}

// curl -X GET "http://localhost:7700/stats"
{
  "databaseSize": 368443392,
  "usedDatabaseSize": 367673344,
  "lastUpdate": "2026-04-16T13:07:02.74243Z",
  "noindexes": {
    "comics": {
      "numberOfDocuments": 31944,
      "rawDocumentDbSize": 24264704,
      "avgDocumentSize": 751,
      "isIndexing": false,
      "numberOfEmbeddings": 0,
      "numberOfEmbeddedDocuments": 0,
      "fieldDistribution": {
        "genres": 31944,
        "id": 31944,
        "overview": 31944,
        "poster": 31944,
        "release_date": 31944,
        "title": 31944
      }
    },
    "movies": {
      "numberOfDocuments": 31944,
      "rawDocumentDbSize": 20594688,
      "avgDocumentSize": 636,
      "isIndexing": false,
      "numberOfEmbeddings": 0,
      "numberOfEmbeddedDocuments": 0,
      "fieldDistribution": {
        "genres": 31944,
        "id": 31944,
        "overview": 31944,
        "poster": 31944,
        "release_date": 31944,
        "title": 31944
      }
    }
  }
}

By @dureuill in #6338

🦋 Fixes

• Settings changes performed via a settings subroute (not the "all settings" route) are now correctly propagated to other remotes on the network. By @dureuill in #6380
• The chat route no longer fails in some condition when using Mistral as a provider. By @renezander030 in #6327
• Prevent accidentally renaming a remote when modifying the network configuration. By @dureuill in #6370

🔩 Miscellaneous changes

• Replace boolean by PatternMatch for searchable by @ManyTheFish in #6344
• Add query to federated search metadata by @StephaneRob in #6371
• Remove issue template for feature issues by @curquiza in #6372
• Improve the documentation of the health route by @Kerollmops in #6376

New Contributors

• @renezander030 made their first contribution in #6327
• @quyentonndbs made their first contribution in #6382

Full Changelog: v1.43.1...v1.44.0

Meilisearch v1.43.1 contains a security fix for an authenticated SSRF vulnerability.

No exploitation was found on Meilisearch Cloud. Cloud users are not required to update.

We recommend that self-hosting users upgrade if they allow third parties to configure Meilisearch instances.

We thank Sion Park (@tldhs1144), who reported the issue and suggested a fix, for improving the security of Meilisearch ❤️

🌈 Enhancements

New settings indexer

The new settings indexer, a more efficient version of our internal indexer for index settings modifications, now supports filterable, sortable, facet search, and custom (asc/desc) attributes in addition to the already-supported searchable, exact, proximity precision, and embedders.

• For Meilisearch Cloud users, the new settings indexer is disabled by default and can be enabled on a case-by-case basis for scaling purposes.
• For OSS users, the new settings indexer can be disabled by setting the MEILI_EXPERIMENTAL_NO_EDITION_2024_FOR_SETTINGS to true.

By @Kerollmops in #6124

Improve facet search indexing performance

We reduced the time to generate the internal data structures for the facet search to work. We were previously doing a couple of full scans on the internal entries and replaced these with scans on specific entries, only the ones dedicated to the facet searchable fields.

By @Kerollmops in #6325

Improve task queue compaction integration

The GET /health route now returns HTTP 500 after a successful task queue compaction, to signal that Meilisearch should be restarted so that tasks can be enqueued again.

For Meilisearch Cloud users, this ensures that compacting the task queue will automatically restart the instance after the compaction.

🦋 Fixes

Fix lexicographic filters on strings

Fixes a bug where string facet values appearing in <, <=, >, >= and IN filters were not normalized before comparison to facet values.

This would cause some values in documents (e.g. 2026-01-01T00:00:00) to appear to be higher than their filter counterpart, due to being normalized (e.g. like 2026-01-01t00:00:00).

Thanks to @njaard for reporting the issue ❤️

By @dureuill in #6340

Fix typo tolerance v1.41 regression

Fixes the WordDelta::added_or_deleted_words function.

By @Kerollmops in #6350.

Thanks to @vladradin99 and @snurfer0 for reporting the issue in #6349 and #6324 ❤️

🔩 Miscellaneous

• Introduce a meilitool command to export an index words FST by @Kerollmops in #6332
• Add riscv64 to release binary matrix via cross-rs by @gounthar in #6296
• Fix docs related to task cancelation and deletion by @curquiza in #6353
• Bump actions/checkout from 5 to 6 by @dependabot[bot] in #6303
• Bump actions/setup-node from 4 to 6 by @dependabot[bot] in #5964
• Bump thin-vec from 0.2.14 to 0.2.16 by @dependabot[bot] in #6339
• Bump docker/setup-buildx-action from 3 to 4 by @dependabot[bot] in #6304
• Bump rustls-webpki from 0.103.10 to 0.103.13 by @dependabot[bot] in #6347
• Bump actix-http from 3.11.2 to 3.12.1 by @dependabot[bot] in #6348
• Bump sigstore/cosign-installer from 3.10.0 to 4.1.1 by @dependabot[bot] in #6301
• Bump supercharge/mongodb-github-action from 1.12.0 to 1.12.1 by @dependabot[bot] in #6302

Thanks again @gounthar, @njaard, @vladradin99 and @snurfer0 for your contributions ❤️

Full Changelog: v1.42.1...v1.43.0

🪲 Bug fixes

Fix issues when using the legacy settings indexer

By @dureuill and @ManyTheFish in #6331

• When using the experimental feature "multimodal", removing a fragment would cause an internal error in subsequent search requests
• When using the experimental feature "multimodal", modifying fragments would sometimes cause an internal error at indexing time
• regenerate: false would be ignored when modifying embedder settings
• When a nested field was declared as searchable, it would not be indexed if its parents were not declared as searchable

🔩 Miscellaneous

• Add new CI run that tests the stable settings indexer by @dureuill in #6331