Vector databasesweaviate.io12 published changes
Users running Weaviate will benefit from improved LSM store performance and fixes to HNSW index implementations.
Users can now provide license keys via file reference (LICENSE_KEY_FILE) in addition to direct environment variable (LICENSE_KEY), improving security and deployment flexibility.
Users seeking dedicated clusters must now contact sales instead of seeing upfront pricing; pricing transparency reduced for dedicated tier.
Users relying on the /v1/modules endpoint, contextionary extensions, or concept handlers will need to update their integration code. This API endpoint is no longer available.
Users can now use DigitalOcean as a generative AI provider within Weaviate
Improves reliability and performance of backup operations, usage tracking, and replication features.
Users can now use TwelveLabs Marengo for multimodal vectorization in Weaviate applications.
Users can now perform more sophisticated BM25 searches across multiple properties with AND matching logic.
Users who deploy Weaviate will receive fixes for range results, namespace handling, backup deduplication, vector index initialization, and performance improvements
Customers on paid tiers will receive a lower service level guarantee, meaning slightly more acceptable downtime in service agreements.
Developers can now use a dedicated REST endpoint for near-text search operations, expanding API capabilities for semantic search use cases.
Users get bug fixes and performance improvements, particularly for LSM operations and batch vectorization which were causing deadlocks.