Experimental features overview

Meilisearch periodically introduces new experimental features. Experimental features are not always ready for production, but offer functionality that might benefit some users. An experimental feature’s API can change significantly and become incompatible between releases. Keep this in mind when using experimental features in a production environment. Meilisearch makes experimental features available expecting they will become stable in a future release, but this is not guaranteed.

Activating experimental features

Experimental features fall into two groups based on how they are activated or deactivated:

Those that are activated at launch with a command-line flag or environment variable
Those that are activated with the /experimental-features API route.

Activating experimental features at launch

Some experimental features can be activated at launch, for example with a command-line flag:

./meilisearch --experimental-enable-metrics

Flags and environment variables for experimental features are not included in the regular configuration options list. Instead, consult the specific documentation page for the feature you are interested in, which can be found in the experimental section. Command-line flags for experimental features are always prefixed with --experimental. Environment variables for experimental features are always prefixed with MEILI_EXPERIMENTAL. Activating or deactivating experimental features this way requires you to relaunch Meilisearch.

Activating experimental features during runtime

Some experimental features can be activated via an HTTP call using the /experimental-features API route:

curl \
  -X PATCH 'MEILISEARCH_URL/experimental-features/' \
  -H 'Authorization: Bearer API_KEY' \
  -H 'Content-Type: application/json'  \
  --data-binary '{
    "metrics": true
  }'

client.update_experimental_features({"metrics": True})

client.update_experimental_features(metrics: true)

client.ExperimentalFeatures().SetMetrics(true).Update()

let client = Client::new("MEILISEARCH_URL", Some("apiKey"));
let features = ExperimentalFeatures::new(&client);
features.set_metrics(true)
let res = features
  .update()
  .await
  .unwrap();

Activating or deactivating experimental features this way does not require you to relaunch Meilisearch.

The logs and metrics experimental features are not available on Meilisearch Cloud. Both require controlling the Meilisearch process at launch to enable the --experimental-enable-logs-route and --experimental-enable-metrics flags, which Cloud users cannot do. Use Cloud’s built-in Analytics dashboard for observability instead.

Current experimental features

Name	Description	How to configure
Limit task batch size	Limits number of tasks processed in a single batch	CLI flag or environment variable
Log customization	Customize log output and set up log streams	CLI flag or environment variable, API route
Metrics API	Exposes Prometheus-compatible analytics data	CLI flag or environment variable, API route
Reduce indexing memory usage	Optimizes indexing performance	CLI flag or environment variable
Search queue size	Configure maximum number of concurrent search requests	CLI flag or environment variable
Drop search after	Drop irrelevant search requests after a configurable timeout (default: 60s)	CLI flag or environment variable
Searches per core	Configure number of concurrent search requests per CPU core (default: 4)	CLI flag or environment variable
`CONTAINS` filter operator	Enables usage of `CONTAINS` with the `filter` search parameter	CLI flag or environment variable, API route
Edit documents with function	Use a Rhai function to edit documents directly in the Meilisearch database	API route
`/network` route	Enable `/network` route	API route
Composite embedders	Enable composite embedders	API route
Chat completions	Enable chat completion capabilities	API route
Get task documents route	Enable route to retrieve documents from tasks	API route
Search query embedding cache	Enable a cache for search query embeddings	CLI flag or environment variable
Maximum batch payload size	Limit batch payload size	CLI flag or environment variable
Multimodal search	Enable multimodal search	API route
Disable new indexer	Use previous settings indexer	CLI flag or environment variable
Allowed IP networks	Override default IP policy with allowed CIDR ranges	CLI flag or environment variable
Search personalization	Enables search personalization	CLI flag or environment variable
Search rules	Curate search results by pinning selected documents when query- or time-based conditions match	API route
Foreign keys	Link documents across indexes to enrich search results with related data and filter by related document properties	API route
Task queue compaction	Compact the task queue database to reclaim space for new tasks	API route
Disable documents fetch queue	Disable the documents fetch queue, which forces document fetch routes to wait in the search queue when no thread is available	API route
Render template	Render document templates and fragments against any input to test embedder configuration	API route
Task and batch streams	Subscribe to live task and batch updates over a stream instead of polling	API route
MCP route	Enable the built-in Model Context Protocol server for search-only AI agent access	API route

Comparisons

Migration

Self-hosting

Demos

Under the hood

Other resources

Activating experimental features

Activating experimental features at launch

Activating experimental features during runtime

Current experimental features

​Activating experimental features

​Activating experimental features at launch

​Activating experimental features during runtime

​Current experimental features

Activating experimental features

Activating experimental features at launch

Activating experimental features during runtime

Current experimental features