Docs Menu
Docs Home
/
MongoDB Atlas
/

Atlas Vector Search Changelog

On this page

  • 2024 Releases
  • 2023 Releases
  • Supports ingesting float32 and int8 vectors using the new BSON vector type.

  • Supports pre-filtering using arrays with all operators.

  • Supports using ObjectId with range operators in Atlas Vector Search pre-filters.

  • Supports using UUID values with all operators in Atlas Vector Search pre-filters.

  • Supports using the $not operator in Atlas Vector Search pre-filters.

  • Upgrades to Java 21. Java 21 supports SIMD, which increases the amount of parallelism the CPU can do during vector comparisons.

    Note

    Atlas will reindex all of your existing Atlas Search indexes on the Atlas cluster in a rolling manner. The Atlas UI will display alerts that indicate the timing of your reindexing. Clusters not using separate search nodes might notice a temporary increase in disk and CPU utilization during reindexing. For clusters using separate search nodes, Atlas will temporarily deploy additional nodes for free for reindexing and there will be no downtime for swapping of indexes when the new index build completes.

  • Introduces exact flag in $vectorSearch to perform ENN search for small-scale, multi-tenant, or recall benchmarking use cases. To learn more, see Run Vector Search Queries.

  • Supports pre-filtering data by date fields indexed as the filter type.

  • Updates Atlas Vector Search queries to run concurrently by default on search nodes.

  • Increases the vector dimension limit to 4096.

  • Makes $vectorSearch generally available for both your development and production deployments. Also makes separate Search Nodes for AWS generally available for both your development and production deployments. To learn more, see Node Architecture.

  • Introduces vectorSearch index type that indexes vector embeddings as vector type for semantic search and boolean, numeric, and string data as filter type for pre-filtering.

Note

We are in the process of rolling out the new Atlas Vector Search experience with a dedicated Atlas Vector Search index definition. It will be available in your project within the next few days or after the next maintenance window.

Introduces $vectorSearch to perform semantic search on data in your Atlas cluster running MongoDB 7.0.2.

Note

Atlas Vector Search is available as a Preview feature. It is made available for evaluation purposes and to gather feedback from users. It is not recommended for production deployments as the feature and the corresponding documentation might change at any time during the Preview period.

To learn more, see Atlas Vector Search Overview.