Weaviate

This is a release candidate (RC) for v1.37.0, featuring several features reaching General Availability (GA) and introducing extensible tokenizers.

This release focuses on stability and security, fixing several bugs related to tenant state management, data filtering, and dependency updates. Key improvements include better handling of INACTIVE tenant backups and security vulnerability patches.

This release candidate (v1.37.0 RC) introduces extensible tokenizers and promotes incremental backups and async replication to General Availability (GA), alongside security hardening.

This release introduces new features like OIDC insecure skip TLS verify and on-demand query profiling, alongside various internal improvements and dependency updates.

This patch release focuses on stability by fixing several bugs, including issues with link reading, compression errors, and improving backward compatibility for module downgrades and dynamic shard loading.

This release introduces token source authentication for GCS backups and usage, alongside numerous bug fixes addressing panics, filtering issues, and performance improvements in backups and compression.

This release focuses heavily on stability, introducing fixes for various race conditions, improving caching, and enhancing the multi2vec-google module with Google AI Studio support and audio capabilities. It also introduces new configuration options for metrics and authentication for GCS operations.

This release focuses primarily on bug fixes, performance improvements, and introducing token source authentication for GCS operations. There are no breaking changes introduced.

This patch introduces audio support for the multi2vec-google module and implements dynamic lazy loading for shards. It also includes numerous stability fixes addressing race conditions, test flakiness, and consistency issues across clustering and backup operations.

This release focuses heavily on stability, fixing numerous race conditions during backup, restore, and general operations. It also introduces support for Google AI Studio models and audio processing within the multi2vec-google module.

This release focuses on stability and improvements, including adding client version to audit logs and enhancing backup and restore reliability. It also introduces support for Google AI Studio models in the multi2vec-google module.

This patch release includes updates to default propagation settings for async replication and a fix ensuring BM25 search does not strip nils.

This patch release includes minor fixes, updating default propagation settings for async replication and ensuring nils are not stripped during bm25 search.

This patch release includes minor fixes, updating default propagation settings for async replication and ensuring BM25 search does not strip nils.

This release introduces file-based incremental backups and adds an RBAC cache lifetime. It also includes various bug fixes, performance improvements, and dependency updates.

This patch introduces file-based incremental backups and enhances stability through various fixes, including improved GRPC error handling and delayed tombstone cleanup. It also updates several security-related dependencies.

This release introduces file-based incremental backups and includes several bug fixes related to debugging, testing, tombstone cleanup, and RBAC caching. Dependency updates for Google Cloud Storage and OpenTelemetry were also performed.

Weaviate v1.36.2 is a patch release focusing on bug fixes, including addressing quantization timing, S3 journey tests, batch additions in EC, and handling zero-length vectors.

This patch release focuses on stability, including a fix to delay quantization until after cache prefilling and ensuring correct handling of zero-length vectors during operations.

This patch release focuses on bug fixes, including delaying quantization until after cache prefilling and ensuring correct handling of zero-length vectors during operations.

This patch release focuses on stability and performance improvements, including reverting a replication retry behavior and optimizing bitmap operations during searches.

This patch release focuses heavily on stability and performance improvements, including better tombstone cleanup, non-blocking segment deletions, and various fixes related to replication and tenant handling.

This patch release includes backported fixes for Claude settings and updates internal indexing logic for vector configurations, along with a fix related to quantization timing.

This release focuses on stability and performance improvements, including better tombstone cleanup, fixes related to replication and Elastic Clustering, and the introduction of non-blocking segment deletions.

This release promotes Server-side Batching, Object TTL, Backup Restore Cancellation, and Dropping Inverted Indices to General Availability (GA), alongside introducing HFresh to Preview with significant performance and feature enhancements.

This release focuses on bug fixes, including improvements to geo filtering and hybrid search reranking, alongside the addition of naive batch logic for the text2vec-google module.

This release focuses heavily on bug fixes, stability improvements, and performance enhancements, including tombstone cleanup optimizations and fixes for flaky tests. New features include batch logic for the text2vec-google module and a debug endpoint for LSM buckets.

This release focuses on stability and performance improvements, including fixing flaky tests, reducing heap usage during compaction, and enhancing logging for panics. A new feature adds recursive property return for nested objects and batch logic to the text2vec-google module.

This release focuses on stability and bug fixes, including removing the use of os.Chdir and addressing potential panics in the transaction manager. A new debug endpoint for LSM buckets has also been introduced.

This release focuses primarily on bug fixes, performance improvements, and stability enhancements, including tombstone cleanup optimizations and fixes for flaky tests.

This release candidate (v1.36.0 RC) marks several features as Generally Available (GA), including Server-side Batching and Object TTL, and introduces the HFresh feature in Preview.

This release introduces the time-based deletion strategy as the default for new collections and includes several fixes related to compression, cluster state management, and migration.

This release introduces the time-based deletion strategy as the default for new collections and includes several fixes related to compression, cluster state management, and data migration.

This patch release focuses on stability and bug fixes, including resolving symlink loops, fixing race conditions during compression, and improving handling of legacy model configurations.

This release focuses on bug fixes, including improvements to async operations, replication stability, and handling of legacy model configurations in text2vec-openai.

This release focuses on stability and bug fixes, including improvements to async operations, fixes in vector storage handling, and compatibility adjustments for text2vec-openai modules.

This patch release includes refactoring of backup acceptance tests and fixes related to backup reporting and data transmission.

This patch release focuses on internal improvements, including refactoring backup acceptance tests and fixing issues related to backup reporting and data transmission.

This patch release primarily addresses a bug fix related to reporting backup sizes.

This patch release focuses on stability, including fixes for WAL recovery, BM25 context checking, and improvements to large shard handling during backups.

This release focuses heavily on performance optimizations across various components, including memtable operations, vector search, and backup handling. It also includes several bug fixes addressing deadlocks and serialization issues.

This release focuses heavily on performance optimizations across memory management, search operations (HNSW re-scoring, vector transfer), and backup handling. Several critical bug fixes address deadlocks and serialization issues.

This release focuses on performance improvements and internal stability fixes, including optimizations in memtable operations and neighbor connection handling, alongside adding cloud provider telemetry.

This patch release includes an optimization to reduce unnecessary network traffic by only sending object merge vectors across the cluster when required.

This release focuses primarily on bug fixes across various components including scheduling, backup operations, HNSW, and distance calculations. It also introduces video modality support for the multi2vec-voyageai module.

This release introduces a bucket load limiter and resolves several bugs related to backup restoration, BMW combined filters, and replication timeouts.

This patch release introduces a bucket load limiter and adds video modality support to the multi2vec-voyageai module, while also fixing an issue related to ResponseHeaderTimeout in HTTP replication.

This release focuses on stability and bug fixes across distance calculation, backup operations, and module handling. Key improvements include better handling of backup status reporting and fixes for various internal inconsistencies.

This release focuses on stability and performance improvements, including fixes for NaN calculations, backup restores, combined filters, and long operation timeouts. A new bucket load limiter has also been introduced.

This release focuses primarily on bug fixes across various components including backup operations, HNSW indexing, and distance calculations. A new feature allows returning the backup size on status requests.

This patch release addresses a bug related to the calculation of average proposal length, preventing NaN values.

This patch release focuses on stability and performance improvements, including fixes for PQ distance calculation and HNSW node handling.

This patch release focuses on bug fixes, including ensuring correct ordering on tied scores and immediate loading of runtime configuration overrides upon startup.

This patch release focuses on bug fixes, including ensuring correct ordering on tied scores and immediate loading of runtime configuration overrides upon startup.

This release focuses primarily on bug fixes and stability improvements, including optimizations to shard handling, fixes for tenant loading, and improvements to runtime configuration loading.

This release focuses on stability and configuration loading, including fixes for SyncShard blocking, Raft error logging, and immediate loading of overridden runtime configurations.

This release focuses on stability and performance improvements, introducing naive batch logic for vectorizer modules and fixing numerous issues related to batch processing, replication, and tenant management.

This patch release focuses on maintenance by removing an outdated model from tests and fixing an issue related to runtime configuration loading and MT sharding state checking.

This release focuses on stability and internal improvements, including optimizations to shard handling, logging adjustments, and fixes related to tenant management and configuration loading.

This release introduces Object TTL for automatic object expiration and major enhancements to the HFresh indexing system (formerly SPFresh). It also brings significant updates to various vectorizer and reranker modules, including VoyageAI v3.5 support and batch processing improvements.

This patch release focuses on improving test stability, updating OIDC configurations, adjusting logging levels, and fixing an issue related to tenant offloading.

This is a release candidate (RC) for v1.35.0, marking feature completion and the end of beta-testing, introducing Operational Modes and Object TTL (Preview).

This patch release focuses heavily on stability, concurrency improvements, and various bug fixes across OIDC configuration, replication handling, backup operations, and internal server logic.

This release focuses on stability and internal improvements, including fixes for OIDC settings, replication handling, and HNSW caching, alongside introducing a maintenance mode interceptor for gRPC.

This release focuses on stability and internal improvements, including fixes for OIDC settings, replication readiness, S3 backup error handling, and various internal component stability enhancements.

This patch release primarily addresses a minor configuration issue related to OIDC runtime settings.

This patch release focuses on minor fixes, including updating proto definitions and ensuring backward compatibility for replication snapshot error parsing.

This patch release primarily focuses on bug fixes, including updating the proto definition for the generative-aws module and ensuring backward compatibility for replication snapshot error parsing.

This patch release includes fixes for the generative-aws module proto definition and backward compatibility improvements for raft replication snapshot error parsing.

This patch release focuses on stability and maintenance, including fixes for race conditions, compaction logic, and improvements to shard loading concurrency. It also backports support for the voyage-3-large module.

This patch release focuses on stability and maintenance, including fixes for race conditions, compaction logic, and improvements to shard loading performance. It also backports support for the voyage-3-large module.

This release introduces support for the voyage-3-large module and includes several stability fixes, notably addressing a remote search race condition and improving shard loading performance.

This release focuses on stability and internal refactoring, including fixes for Gemini API batch requests, safer file downloads, and configuration debugging endpoints.

This patch release focuses on fixing issues related to the Gemini API batch requests and batch text vectorizer handling of empty objects, and introduces a new debug endpoint for configuration inspection.

This patch release includes fixes for the Gemini API batch requests and batch text vectorizer handling of empty objects, alongside a new debug endpoint for configuration inspection.

This patch release focuses heavily on stability, CI improvements, and module enhancements, including support for VoyageAI v3.5 models and batch API improvements for text2vec modules. Several fixes address issues related to replication, testing flakiness, and backup operations.

This release introduces an internal gRPC server equivalent to the REST clusterapi and includes numerous fixes related to debugging endpoints, module batching, backup operations, and error handling across various modules.

This release introduces new debug endpoints for memory and CPU limits, adds zstd compression for backups, and includes various fixes related to module batching, error logging, and S3 uploads.

This release focuses on stability, performance improvements across various subsystems like latency reduction and replication handling, and introduces support for new embedding models and Geo HNSW configuration.

This release focuses heavily on bug fixes, performance improvements, especially around latency and replication scaling, and adds support for new embedding models and Geo HNSW configuration.

This release focuses heavily on performance improvements, particularly around latency reduction and replication scaling. It also introduces new features for Geo HNSW configuration and updates support for various generative AI models.

This minor release focuses on improving internal stability by enhancing scratch space cleanup mechanisms.

This release introduces significant new features including Server-side Dynamic Batching and the SPFresh vector index, alongside performance improvements and dependency security updates. It also adds support for new generative AI modules and quantization methods.

This patch release focuses on improving the reliability and error handling of the WriteTo method, specifically concerning scratch space management.

This patch release focuses primarily on bug fixes, including addressing resource leaks, stopword update issues, and various path fixes. It also introduces an environment variable for controlling replica movement.

This release introduces support for Multi DC configurations by separating advertise and bind addresses and includes several bug fixes related to resource management, stability tests, stopwords, and backup paths.

This patch release focuses on stability, addressing several issues related to RAFT state management, waitgroup handling, and backup paths.

This release candidate (v1.34.0 RC) introduces server-side dynamic batching (Preview) and support for Flat index RQ quantization.

This release focuses primarily on stability and bug fixes, including improvements to e2e testing infrastructure, resource management, and specific fixes related to stopwords and backup paths.

This patch release focuses on stability and performance improvements, including fixes for data races, reduced memory usage, and reworked LSM Store synchronization. A key feature is the separation of advertise and bind addresses for multi-DC support.

This release focuses on stability and performance improvements, including fixes for data races, reduced memory usage, and reworked synchronization in the LSM Store. A new feature allows separation of advertise and bind addresses for multi-DC support.

This patch release focuses heavily on stability and performance improvements, addressing several data races, improving test reliability, and reworking LSM Store synchronization to reduce latency spikes.

This patch release focuses primarily on bug fixes across various components including data race conditions, RAFT state restoration, replication tests, and quantization vector handling. It also introduces the ability to sort the backup list and return backup size.

This release focuses primarily on stability and bug fixes across raft, replication, and backup testing, while introducing the ability to sort the backup list and return backup size.

This patch release addresses two specific bugs: a data race in the commit logger and an issue related to RQ quantization vectors.

This release candidate (v1.34.0 RC) introduces server-side dynamic batching (Preview) and support for Flat index RQ quantization.

This release focuses heavily on stability and performance improvements, particularly around RAFT consensus, memberlist configuration, and extensive additions to metrics collection across various components like LSM, compaction, and replication. It also introduces image support for the generative-cohere module.

This release focuses on stability and bug fixes, including improvements to logging, handling of compressed vectors, and migration logic for vector indices. A new feature adds 1% sampled queries to the slow log for better performance monitoring.

This release focuses heavily on stability, particularly around RAFT consensus, replication, and LSM storage performance, introducing numerous bug fixes and significant metric enhancements. Key changes include improved handling of failed broadcasts, better RAFT memberlist configuration, and the addition of extensive new metrics across LSM components.

This release focuses heavily on stability and bug fixes, particularly around vector indexing, compression handling, and migration processes involving legacy and named vectors. Several race conditions and panics have been resolved.