Qdrant

Qdrant releases performance optimizations for AVX systems and BM25 search

Qdrant released performance improvements for dot product calculations on AVX systems and fixed BM25 search functionality issues. The update also includes Rust version management improvements in the Nix package. These are maintenance updates that improve system performance and reliability without requiring user action.

Date not specified

Qdrant vector database improves BM25 inference and performance

This release adds BM25 inference capabilities and improves performance for mutable map indexes used in full text search. Multiple consistency and data integrity issues were resolved, along with UI enhancements for cluster visualization.

Date not specified

Qdrant fixes critical shard routing and UUID index issues in v1.15.1

Multiple bug fixes addressing critical issues including broken point shard routing in version 1.15 that could cause improper behavior on updates, and UUID index storage problems resulting in missing filter results. Users running version 1.15.0 with multi-shard collections should review their point data integrity and consider updating immediately.

Date not specified

Qdrant releases v1.15.0: Multilingual support and quantization improvements

Qdrant has released significant enhancements to their vector database including new multilingual tokenization, advanced quantization methods (2-bit and 1.5-bit binary), phrase matching, and various performance optimizations. These updates improve search capabilities and efficiency but do not require immediate action from existing users. The changes represent substantial improvements to the platform's search and indexing capabilities.

Date not specified

Qdrant Performance Improvements: Indexing, Storage, and Query

Multiple performance enhancements including faster WAL-delta transfers, improved GPU indexing, optimized payload index loading, and better resource scheduling. These updates primarily improve system performance and efficiency without breaking existing functionality. No action required from users as these are backend optimizations.

Date not specified

Qdrant Vector Database Release v1.14.0: Score Boosting, Recommendations, Performance

This release introduces server-side score boosting with user-defined formulas and a new sum_scores recommendation strategy for relevance feedback. Additional improvements include incremental HNSW building, better memory optimization, and various bug fixes for crash prevention and data consistency. These are additive features that enhance existing functionality without breaking changes.

Date not specified

Qdrant query API performance improvements released

Multiple performance optimizations to the query API including reduced network overhead and improved search performance with multiple segments. Also includes fixes for resharding transfers and order_by functionality. These are backend improvements that enhance system performance without requiring user action.

Date not specified

Qdrant: Performance Optimizations & Critical Cluster Corruption Fixes

This release includes significant performance improvements for CPU/IO optimization, query performance, and GPU operations. Multiple critical fixes address potential cluster corruption issues in consensus snapshots and user-defined sharding that could cause data integrity problems. Users should update promptly to avoid cluster state corruption, especially those using user-defined sharding or consensus snapshots.

Date not specified

Qdrant fixes critical deadlock and HNSW index bug

This release addresses a potential deadlock in consensus logic that could cause nodes to become stuck, along with fixing HNSW index on-disk flag behavior and replica transfer issues. Users should update to prevent potential system stability problems, particularly in clustered deployments.

Date not specified

Qdrant database improvements: consensus compaction, data consistency, and performance

This release introduces multiple database improvements including enabled consensus compaction by default for faster peer operations, enhanced data consistency by excluding unversioned points from reads/writes, and various performance optimizations. The updates also include REST API improvements, rate limiting enhancements, and several bug fixes for system stability.

Date not specified

Qdrant improves GPU compatibility and fixes data handling issues

This update adds fallback support for GPUs without half-float precision and fixes several stability issues including blob store panics and optimizer data resilience problems. These are maintenance improvements that enhance system reliability without requiring user action.

Date not specified

Critical Bug Fixes Released for GPU HNSW, Memory Management, and Data Persistence

This release addresses several serious issues including potential panics in GPU HNSW operations, memory mapping problems, and data persistence failures. Users should update promptly to avoid crashes and data loss, particularly those using GPU acceleration or custom sharding configurations.

Date not specified

Qdrant adds GPU support and performance optimizations

Qdrant has released significant performance improvements including GPU support for faster HNSW indexing, runtime resharding capabilities, and optimized memory-mapped storage for payloads and sparse vectors. The update also includes new vector filtering conditions, strict mode for collection operations, and various bug fixes. These are additive features that enhance performance and functionality without breaking existing implementations.

Date not specified