New Features
Approximate facet distribution with sampledCount sort
Facet distribution now supports a new sampledCount sort option. It gives a faster, approximate estimate of the facet distribution than the count sort. The document count for each facet is exhaustive, but the list of facets is not: some facets that the count sort would list may be skipped.
Use sampledCount when you need fast facet results and can accept an incomplete facet list.
Improvements
Adaptive rate limiting for embedders
Meilisearch now uses an adaptive strategy to handle embedder load and outages:- For small workloads with few documents, Meilisearch accepts slower embedder services
- For large updates with thousands of documents, Meilisearch gives up early to avoid long delays
- If the embedder server is unavailable for too long, Meilisearch abandons indexing early
- Meilisearch respects the
Retry-Afterheader (in seconds) to avoid unnecessary retries during brief service interruptions