← Back to all sparks
M

Meilisearch

DEVOPS
Velocity5.0

Open-source, fast, typo-tolerant search engine API.

Meilisearch pushes indexing speed and hardens its distributed enterprise tier

searchindexing-performancevector-searchshardinghigh-availabilityopen-source
Current state
Meilisearch is shipping a tight stream of point releases focused on indexing performance and stability. The new settings indexer keeps gaining parameters and speed (v1.46.0), while a run of 1.45.x releases cleaned up a deletion-batching regression and a vector-store corruption bug affecting embedder upgrades. Underneath the patches, the enterprise sharding and replication feature gained a remote-availability fallback.
Where it's heading
The arc is two-track: continuous performance work on the core indexer, and a maturing enterprise distributed story spanning sharding, replication, and high-availability fallback. Vector and embedder support remain an active, still-stabilizing surface.
Prediction
Expect continued settings-indexer performance work and further hardening of the embedder/vector path, with the distributed enterprise features accruing more resilience tooling.

Recent moves

  1. 5h ago

    v1.46.0: faster settings indexer, deletion-batching fix

    Extends the new settings indexer with exact-words and disable-on-words support for faster indexing, and fixes a v1.45.0 regression in how deletions-by-filter were batched with other writes.

    View source ↗
  2. 6d ago

    v1.45.2: fix corrupted embeddings on dumpless upgrade

    Repairs a vector-store quantization config desync that corrupted embeddings when dumpless-upgrading databases from pre-v1.33.1, reinforcing the still-stabilizing embedder path.

    View source ↗
  3. 10d ago

    v1.45.1: revert autobatch deletions-by-filter

    Reverts the v1.45.0 autobatching of deletions-by-filter with additions after it proved problematic, a quick stability correction within the 1.45 line.

    View source ↗
  4. 13d ago

    v1.45.0: faster settings indexing and document fetch

    Improves indexing performance when changing settings and speeds up document fetch, continuing the core indexer optimization theme that runs through this whole window.

    View source ↗
  5. 1mo ago

    v1.42.0: remote-availability fallback for sharding/replication

    ⚡ SPARK

    Adds a fallback system to the enterprise sharding and replication feature that detects unavailable remotes, routes around them, and resumes once they recover, exposed via the /network route. This is the clearest move in the window toward a resilient distributed Meilisearch.

    View source ↗
  6. 1mo ago

    v1.42.1: legacy settings indexer and multimodal fixes

    Fixes several legacy settings-indexer and experimental multimodal embedder errors, including fragment edits causing internal errors and regenerate:false being ignored, tightening the embedder surface.

    View source ↗