Weaviate — The AI Toolchain

Weaviate v1.38.13

Thu, 27 Aug 2026 10:13:43 GMT

Weaviate v1.38.13 adds a DigitalOcean generative module and export/import API endpoints for database user API-key hashes. • Adds export and import API endpoints for database user API-key hashes, enabling backup and migration of user credentials. • Adds a DigitalOcean generative module, extending the set of supported generative AI providers. • MCP server now runs in stateless streamable mode and correctly refuses GET requests with HTTP 405.

Weaviate v1.37.15

Thu, 27 Aug 2026 06:41:32 GMT

Weaviate v1.37.15 adds TwelveLabs Marengo multimodal vectorizer, DigitalOcean generative module, cross-property AND matching in BM25, and parallelized rescoring. • Adds `generative-digitalocean` module for using DigitalOcean's generative AI models as a Weaviate generative backend. • Adds cross-property AND matching in BM25 search, allowing terms to be required across multiple properties simultaneously. • Adds per-query concurrency budget enforcement in HNSW compressed rescore via `feat(hnsw): respect per-query concurrency budget in compressed rescore`. • Parallelizes HNSW Muvera late-interaction rescoring for improved throughput on multi-vector workloads. • Parallelizes BM25 block term creation across properties, reducing indexing latency at scale. • Parallelizes `hfresh` rescoring with budget-aware workers and pooled buffer reads for fresh-index queries. • Adds MUVERA-specific usage calculations for accurate billing and resource tracking on multi-vector collections. • Allocates per-tenant vector cache memory lazily and proportionally to tenant size, reducing idle memory overhead in multi-tenant deployments. • Warms the `hfresh` version map in the background at startup, reducing cold-start latency for fresh-index queries. • Adds targeted replace scan with newest-wins visibility in `lsmkv`, improving read performance on frequently updated keys. • Registers `disable_dimension_metrics` as a runtime override, allowing it to be toggled without a restart.

Weaviate v1.39.2

Wed, 26 Aug 2026 17:21:38 GMT

• New `generative-digitalocean` module adds DigitalOcean as a generative AI provider.

Weaviate v1.39.1

Tue, 25 Aug 2026 08:35:33 GMT

Weaviate v1.39.1 adds new REST Search API endpoints, backup role inclusion, GCS gRPC transport, and auto-schema named vector defaults. • Adds `includeRoles` parameter to the backup/restore process to include role definitions in backups. • Adds BM25 keyword search REST endpoint (`POST /v1/search/bm25` or equivalent REST Search API 1/4). • Adds hybrid search REST endpoint (REST Search API 2/4). • Adds near-object search REST endpoint (REST Search API 3/4). • Adds aggregate counts REST endpoint (REST Search API 4/4). • Adds opt-in gRPC transport for the `backup-gcs` backend. • Introduces `RUNTIME_REINDEX_ENABLED` kill-switch environment variable (off by default). • Auto-schema now creates a `default` named vector instead of a legacy vector when inferring schema. • Adds MUVERA-specific usage calculations for multi-vector index accounting. • Resumes interrupted vector-index drop operations from the recorded pending set, improving reliability of drop-vector-index across restarts.

Weaviate 1.39.0

Mon, 24 Aug 2026 15:05:02 GMT

API surface changed: +4 endpoints, 1 modified • New endpoint POST /aggregate/{collection} • New endpoint POST /search/{collection}/bm25 • New endpoint POST /search/{collection}/hybrid • New endpoint POST /search/{collection}/near-object • POST /schema/{className}/properties/{propertyName}/index/{indexName}/cancel: response schema changed

Weaviate 1.39.0

Thu, 20 Aug 2026 10:02:00 GMT

Weaviate now publishes an API — 119 endpoints across 20 areas: Schema, Objects, Authz, … • Schema (26 endpoints) — Operations related to managing collections. • Objects (19 endpoints) — Operations for managing individual data objects. • Authz (18 endpoints) — Endpoints for managing Weaviate's Role-Based Access Control (RBAC) system. • Replication (10 endpoints) — Operations related to managing data replication, including initiating and monitoring shard replica movements between nodes, querying current sharding states, and managing the lifecycle of replication tasks. • Users (8 endpoints) — Endpoints for user account management in Weaviate. • Backups (7 endpoints) — Operations related to creating and managing backups of Weaviate data. • Namespaces (7 endpoints) — Operations for managing cluster-level namespaces. • Batch (3 endpoints) — Operations for performing actions on multiple data items (objects or references) in a single API request. • 12 more areas: Export, Mcp, Well Known, Classifications, Graphql, Nodes, Cluster, Distributedtasks, Meta, Root, Search, Tokenize

Weaviate v1.38.10

Tue, 18 Aug 2026 07:42:19 GMT

Weaviate v1.38.10 adds REST Search API endpoints for BM25, hybrid, near-object, and aggregate-counts queries, plus `includeRoles` in backup/restore and opt-in gRPC transport for GCS backups. • Adds `includeRoles` parameter to the backup and restore process, enabling role definitions to be captured and replayed alongside data. • Adds opt-in gRPC transport for the `backup-gcs` module, available as a new configuration option on the GCS backup provider. • Adds a BM25 keyword search REST endpoint as part of the new REST Search API (`feat(rest): bm25 keyword search endpoint`). • Adds a hybrid search REST endpoint as part of the new REST Search API (`feat(rest): hybrid search endpoint`). • Adds a near-object search REST endpoint as part of the new REST Search API (`feat(rest): near-object search endpoint`). • Adds an aggregate counts REST endpoint as part of the new REST Search API (`feat(rest): aggregate counts endpoint`).

Weaviate v1.38.10

Tue, 18 Aug 2026 07:42:19 GMT

Weaviate v1.38.10 adds REST Search API endpoints for BM25, hybrid, near-object, and aggregate searches, plus `includeRoles` in backup/restore and opt-in gRPC transport for GCS backups. • Adds `includeRoles` option to the backup and restore process, enabling role definitions to be captured and replayed with a backup. • Adds opt-in gRPC transport for the `backup-gcs` module, improving throughput for GCS-backed backups. • Adds REST BM25 keyword search endpoint (REST Search API 1/4), exposing a dedicated REST surface for keyword search. • Adds REST hybrid search endpoint (REST Search API 2/4), exposing a dedicated REST surface for hybrid (vector + keyword) search. • Adds REST near-object search endpoint (REST Search API 3/4), exposing a dedicated REST surface for vector similarity search by object reference. • Adds REST aggregate counts endpoint (REST Search API 4/4), exposing a dedicated REST surface for aggregate count queries.

Weaviate v1.38.9

Thu, 06 Aug 2026 13:57:10 GMT

Weaviate v1.38.9 adds TwelveLabs Marengo multimodal vectorizer and a kill switch for runtime reindexing. • Adds `RUNTIME_REINDEX_ENABLED` environment variable as a kill switch for runtime reindexing (off by default). • Adds `multi2vec-twelvelabs` vectorizer module integrating the TwelveLabs Marengo multimodal embedding model. • Parallelizes HNSW Muvera late-interaction rescoring and budget-aware rescore workers, unlocking higher-throughput ANN search at scale. • Parallelizes BM25 block term creation across properties, improving indexing performance for multi-property collections.

Weaviate v1.38.9

Thu, 06 Aug 2026 13:57:10 GMT

Weaviate v1.38.9 adds TwelveLabs Marengo multimodal vectorizer and a per-query concurrency budget for compressed HNSW rescoring. • Adds `RUNTIME_REINDEX_ENABLED` environment variable kill switch (off by default) to control runtime reindexing. • Adds `multi2vec-twelvelabs` vectorizer module integrating TwelveLabs Marengo for multimodal vectorization. • Adds per-query concurrency budget enforcement in compressed HNSW rescore operations. • Parallelizes BM25 block term creation across properties, unlocking higher indexing throughput. • Parallelizes HNSW Muvera late-interaction rescoring for faster approximate nearest-neighbor queries. • Parallelizes `hfresh` rescoring with budget-aware workers and pooled buffer reads, improving query performance.

Weaviate v1.39.0

Tue, 04 Aug 2026 11:42:14 GMT

Weaviate v1.39.0 adds gRPC-web, REST search, 4-bit RQ, Hybrid MMR, cross-property BM25 AND matching, drop-vector-index, and namespace suspend endpoints. • Adds `POST /v1/search/{collection}/near-text` REST endpoint for near-text search queries, with responses enveloped as `{id, properties, references, metadata}` and camelCase payload fields with a nested `rerank` object. • Introduces `/grpc-web` endpoint, enabling gRPC-web protocol access to Weaviate. • Adds GA resource-oriented index endpoints for the Alter Schema reindex feature (v1.39 RFC rework). • Adds namespace suspend endpoints, RAFT state management, and DB-user status checks for namespace lifecycle control on shared clusters. • Returns the first letters of API keys to admins on namespaced clusters. • Adds namespace graduation via backup/restore. • Adds namespace-local roles for per-namespace RBAC isolation. • Introduces drop-vector-index capability: supports removing a vector index from an existing collection property to reclaim disk space, with RBAC integration, multi-tenancy support, and cold-tenant completion. • Adds cross-property AND matching in BM25 search, allowing queries to require term matches across multiple properties simultaneously. • Adds 4-bit Rotational Quantization (RQ4) with improved SIMD vector search performance. • Adds Maximal Marginal Relevance (MMR) support in Hybrid queries for result diversity. • Adds soft-ranking Boost API with missing-property handling and property-type validation. • Parallelizes block term creation across properties in BM25, improving indexing throughput. • Multiple BM25/BlockMax WAND performance optimizations: faster varint decoding, reduced hot-path allocations, tiered merged filter, approximate IDF object count, and deferred tombstone checks.

Weaviate v1.39.0

Tue, 04 Aug 2026 11:42:14 GMT

Weaviate v1.39 adds gRPC-web, REST search, BM25 cross-property AND, 4-bit RQ, MMR hybrid, and namespace suspend endpoints. • Introduces `POST /v1/search/{collection}/near-text` REST endpoint for near-text search queries, with response envelope containing `id`, `properties`, `references`, and `metadata` fields and camelCase payload fields. • Introduces `/grpc-web` endpoint, enabling gRPC-web protocol access to the Weaviate API. • Adds 4-bit Rotational Quantization (RQ4) with improved SIMD vector search performance. • Adds namespace suspend endpoints to the control plane for suspending namespaced tenants, including RAFT state suspension and DB User status checks. • Adds support for cross-property AND matching in BM25 search. • Adds Maximal Marginal Relevance (MMR) support in Hybrid queries for diversity-aware result ranking. • Adds GA resource-oriented index endpoints for the Alter Schema reindex API (v1.39 RFC rework). • Adds drop-vector-index capability (preview): removes inverted/vector indices from existing properties to reclaim disk space, with RBAC integration, multi-tenancy support, and cold-tenant completion. • Returns first letters of API keys to admins on namespaced clusters. • Adds namespace local roles, allowing per-namespace RBAC role scoping. • Adds namespace graduation via backup/restore workflow. • Adds gate to disallow global non-operator users and denies operator-only surfaces to namespaced users. • Delivers multiple BM25/BlockMax WAND hot-path performance improvements including parallelized block term creation, tiered merged filters, approximate IDF object counts, and tombstone/filter probe optimizations.

Weaviate v1.38.8

Wed, 29 Jul 2026 16:57:51 GMT

Weaviate v1.38.8 adds cross-property AND matching in BM25, new S3 auth broker credentials, and a runtime-overridable batched Contains gate. • Adds feat(backup-s3) new auth broker credentials for the S3 backup module, enabling alternative credential flows beyond static keys. • Adds a runtime-overridable feature gate to opt into batched `ContainsAny`/`ContainsAll`/`ContainsNone` resolution via the `config, inverted` layer. • Adds cross-property AND matching support in BM25 search, allowing BM25 queries to require match terms across multiple properties simultaneously. • Exposes the first letters of API keys to admins on namespaced clusters, improving key identification without revealing secrets. • Adds nested object filtering support to the usage module.

Weaviate v1.38.8

Wed, 29 Jul 2026 16:57:51 GMT

Weaviate v1.38.8 adds cross-property AND matching in BM25, new S3 auth broker credentials, and nested object filtering in the usage module. • Adds a runtime-overridable feature gate to enable batched `Contains` (opt-in) for inverted-index queries via a config/runtime override. • Adds cross-property AND matching support in BM25 search, expanding keyword-search relevance control. • Introduces new auth broker credentials for the S3 backup module (`backup-s3`), enabling additional authentication methods. • Returns the first letters of API keys to admins on namespaced clusters, improving key auditability. • Adds nested object filtering support to the usage module. Breaking changes: • Support for restoring old backup formats has been removed; backups created in legacy formats can no longer be restored.

Weaviate v1.37.14

Mon, 27 Jul 2026 14:38:00 GMT

Weaviate v1.37.14 adds unified background-process metrics, persistent cluster identity, and configurable incremental-backup deduplication. • Makes the number of files deduplicated in incremental backups configurable. • Adds unified background-process activity and duration metrics via feat(monitoring) instrumentation. • Adds persistent cluster and node identity for correlatable telemetry across restarts. • Replaces the per-tick due-scan in the cycle manager with a due-heap scheduler, reducing CPU overhead.

Weaviate v1.37.14

Mon, 27 Jul 2026 14:38:00 GMT

Weaviate v1.37.14 adds unified background-process metrics, persistent cluster identity, and configurable incremental backup deduplication. • Makes the number of files deduplicated in incremental backups configurable. • Adds unified background-process activity and duration metrics via feat(monitoring) instrumentation. • Adds persistent cluster and node identity for correlatable telemetry across restarts. • Replaces per-tick due-scan with a due-heap scheduler in cyclemanager for more efficient background task dispatch. • Improves segment index performance using a van Emde Boas layout.

Weaviate v1.38.5

Thu, 16 Jul 2026 18:45:15 GMT

Weaviate v1.38.5 adds a structured where-filter to the MCP hybrid search tool and warns when usage collection cycles overlap. • Exposes a structured where-filter on the MCP hybrid search tool, enabling filtered hybrid search through the MCP interface. • Adds a warning when usage collection cycles overlap the configured collection interval, surfacing capacity issues in usage accounting.

Weaviate v1.38.3

Fri, 10 Jul 2026 07:12:44 GMT

Weaviate v1.38.3 adds a `/grpc-web` endpoint, a runtime GraphQL toggle, hard-link replica movement, and namespace-local RBAC roles. • Introduces the `/grpc-web` endpoint, enabling gRPC-Web protocol support for browser and proxy-constrained clients. • Adds a runtime toggle for the GraphQL API, allowing operators to enable or disable the GraphQL surface without restarting the node. • Adds namespace-local roles, scoping RBAC role definitions to individual namespaces rather than globally. • Adds a gate to disallow global non-operator users, restricting operator-only surfaces from namespaced users. • Supports `automaxprocs` for automatic `GOMAXPROCS` tuning via cgroup v2, improving CPU scheduling in containerized deployments. • Propagates raw on-disk object bytes in async replication, reducing serialization overhead during replica sync. • Uses batched hashtree-root pre-filtering for many-tenant clusters in async replication, reducing per-hashBeat overhead. • Increases the `hfresh` `searchProbe` default to 256, improving recall for hybrid-fresh index queries. • Optimizes the cycle manager for large multi-tenant collections, reducing overhead when managing many tenants.

Weaviate v1.38.2

Thu, 25 Jun 2026 06:02:05 GMT

Weaviate v1.38.2 adds generative-deepseek module, location/endpoint/dimensions settings for Google, OpenAI, and AWS modules. • Adds `generative-deepseek` module with support for a `stop` setting in module settings. • Adds `location` configuration setting to the `text2vec-google` module. • Adds `location` setting support to the `generative-google` module. • Adds `endpoint` setting support in the OpenAI client module. • Adds `dimensions` setting support to the `text2vec-aws` module. • Validates `X-*-BaseURL` request headers across modules to close an SSRF bypass vector. • Increases the default `searchProbe` value to 256 for improved hfresh search behavior.

Weaviate v1.37.10

Wed, 24 Jun 2026 13:26:02 GMT

Weaviate v1.37.10 adds generative-deepseek module, OpenAI endpoint setting, Google location support, and AWS dimensions setting. • Adds `generative-deepseek` module with support for a `stop` setting in module settings. • Adds `endpoint` setting to the OpenAI client module configuration. • Adds `location` setting to the `generative-google` module. • Adds `location` configuration to the `text2vec-google` module. • Adds `dimensions` setting to the `text2vec-aws` module. • Validates `X-*-BaseURL` request headers to close an SSRF bypass vector in modules. • Increases the `searchProbe` default to 256 for hfresh vector index searches.

Weaviate v1.36.19

Wed, 24 Jun 2026 13:11:49 GMT

Weaviate v1.36.19 adds the generative-deepseek module, location and endpoint settings for Google/OpenAI/AWS modules, and SSRF header validation. • Adds `generative-deepseek` module with support for a `stop` setting in module configuration. • Adds `location` configuration setting to the `text2vec-google` module. • Adds `location` setting support to the `generative-google` module. • Adds `endpoint` setting support in the OpenAI client module configuration. • Adds `dimensions` setting support in the `text2vec-aws` module. • Validates `X-*-BaseURL` request headers in modules to close an SSRF bypass vector.

Weaviate v1.38.0

Fri, 05 Jun 2026 16:34:27 GMT

Weaviate v1.38 adds Namespaces (Preview), Nested Object Filtering (Preview), Runtime Property Reindex (Preview), and promotes HFresh to GA. • Introduces Namespaces (Preview) — control-plane and data isolation between users on a shared cluster, with RBAC/OIDC wiring, per-namespace collection limits, object limits, cascading delete, user management, alias endpoints, and audit log entries. • Adds Nested Object Filtering (Preview) — enables search and filtering within indexed JSON properties, supporting `IsNull`, positional `arr[N]` filtering, `Contains*` operators, correlated AND resolution, scope-aware NOT, and gRPC + GraphQL ingress for nested filter paths. • Adds Runtime Property Reindex (Preview) — allows changing a property's index type at runtime without recreating the collection, with two-phase RAFT swap barrier for semantic migrations and graceful-restart resilience for in-flight reindex units. • Promotes HFresh index to GA, with asymmetric distance computation, query-vector normalization before rescoring, and reduced posting-map memory usage.

Weaviate v1.37.5

Tue, 26 May 2026 09:48:53 GMT

Weaviate v1.37.5 adds a DigitalOcean text embedding module, named vector support in the default vector index, and vector index compression allow-lists. • Adds `text2vec-digitalocean` module for generating text embeddings via DigitalOcean's embedding service. • Supports named vectors in the default vector index, enabling multi-vector configurations without specifying a custom index per vector. • Adds allow-lists for vector index compression, giving operators fine-grained control over which vectors are subject to compression. • Adds validation for reserved property name suffixes in schema definitions, preventing naming conflicts at collection creation time.

Weaviate v1.36.15

Fri, 22 May 2026 12:27:08 GMT

Weaviate v1.36.15 adds a new text2vec-digitalocean vectorization module. • Adds `text2vec-digitalocean` module, enabling DigitalOcean-hosted embedding models as a vectorization source.

Weaviate v1.35.21

Thu, 21 May 2026 20:04:15 GMT

Weaviate v1.35.21 adds a new `text2vec-digitalocean` vectorization module. • Adds the `text2vec-digitalocean` module, enabling vectorization of text using DigitalOcean's embedding models as a new integration.

Weaviate v1.37.2

Thu, 23 Apr 2026 20:56:09 GMT

Weaviate v1.37.2 speeds up collection export snapshots and adds asymmetric distance computation to hfresh. • Speeds up collection export snapshotting via concurrent tenant de-activation, reducing snapshot time for multi-tenant collections. • Adds asymmetric distance computation to the `hfresh` index, improving approximate nearest-neighbor search accuracy for quantized vectors.

Weaviate v1.37.0

Thu, 16 Apr 2026 13:58:51 GMT

Weaviate v1.37.0 adds a native MCP server, BlobHash property type, collection export, extensible tokenizers, and drop-vector-index support. • Adds `BlobHash` property type that automatically stores blob data as hashes, greatly reducing disk space required for blob workloads. • Introduces a native MCP (Model Context Protocol) server interface (preview), enabling AI agents such as Claude and IDEs to natively read and write to Weaviate without custom code, with hybrid search, RAG, and multi-tenancy support out of the box. • Adds collection export to cloud backends (AWS, GCP, Azure) and filesystem with point-in-time exports, multi-node support, concurrent exports, multi-tenancy handling, cancellation, and observability; disabled by default and configured via environment variables including a default path env var. • Adds a tokenizer endpoint and middleware integration for extensible tokenizers (Phase 1), bringing self-serve, multilingual tokenization with accent-insensitive processing options for text properties and custom stopword presets. • Adds experimental support for dropping vector indices from existing collections via an alter-schema endpoint, allowing memory reclamation; endpoint can be disabled via an environment setting. • Adds file-based incremental backups, chunked backup file splits for large collections, backup/restore of `INACTIVE` tenants, and avoids halting compactions during backup. • Migrates replica internal cluster communication from REST to gRPC for improved performance and security hardening. • Adds Google AI Studio model support and audio support to the `multi2vec-google` module. • Introduces token source authentication for `backups-gcs` and `usage-gcs` modules. • Adds S3 assume-role support for S3-backed storage. • Adds HFresh index preview improvements including an auto category, increased max posting size floor, and continued operation during backups. • Adds validation that async replication is enabled before use, with async replication now production-ready.

Weaviate v1.35.17

Wed, 15 Apr 2026 17:42:00 GMT

Weaviate v1.35.17 adds backup/restore for INACTIVE tenants and Google AI Studio API key support in multi2vec-google. • Adds support for Google AI Studio API key headers in the `multi2vec-google` module. • Adds `baseURL` validation support. • Adds support for `compactv2` downgrades in HNSW. • Provides a descriptive error on downgrade paths when a module is not available (backward-compatibility improvement).

Weaviate v1.36.9

Fri, 03 Apr 2026 06:56:04 GMT

Weaviate v1.36.9 adds on-demand query profiling and implements AUTHENTICATION_OIDC_INSECURE_SKIP_TLS_VERIFY support. • Implements `AUTHENTICATION_OIDC_INSECURE_SKIP_TLS_VERIFY` environment variable to allow skipping TLS verification for OIDC authentication. • Adds on-demand query profiling support for runtime performance inspection of queries.

Weaviate v1.35.16

Thu, 26 Mar 2026 07:22:37 GMT

Weaviate v1.35.16 adds token source authentication for GCS backup and usage integrations. • Introduces token source authentication support for the `backups-gcs` and `usage-gcs` integrations, enabling credential-less auth flows for GCS-backed operations.

Weaviate v1.34.20

Thu, 26 Mar 2026 07:21:37 GMT

Weaviate v1.34.20 adds Google AI Studio support, audio vectorization, GCS token-source auth, and a new dimension-metrics control flag. • Adds `DisableDimensionMetrics` configuration to control whether dimension metrics are reported. • Introduces token source authentication for the `backups-gcs` and `usage-gcs` modules. • Adds audio support to the `multi2vec-google` module for multimodal vectorization. • Adds support for Google AI Studio models in the `multi2vec-google` module.

Weaviate v1.36.6

Thu, 19 Mar 2026 18:42:52 GMT

Weaviate v1.36.6 adds audio support to multi2vec-google, a DEFAULT_SHARDING_COUNT env var, and a DisableDimensionMetrics config option. • Adds `DEFAULT_SHARDING_COUNT` environment variable to override the default shard count at the instance level. • Adds `DisableDimensionMetrics` configuration option to control whether dimension metrics are reported. • Adds audio modality support to the `multi2vec-google` module for multimodal vectorization. • Adds IPv6 support for cluster networking. • Enables dynamic lazy loading of shards to improve startup and resource utilization.

Weaviate v1.35.15

Thu, 19 Mar 2026 13:20:39 GMT

Weaviate v1.35.15 adds audio support to multi2vec-google, a new DEFAULT_SHARDING_COUNT env var, and a DisableDimensionMetrics config option. • Adds `DEFAULT_SHARDING_COUNT` environment variable to override the default shard count at the server level. • Adds `DisableDimensionMetrics` configuration option to control whether dimension metrics are reported. • Adds audio support to the `multi2vec-google` module, expanding multimodal vectorization beyond text and images. • Adds support for Google AI Studio models in the `multi2vec-google` module.

Weaviate v1.36.5

Thu, 12 Mar 2026 09:19:30 GMT

Weaviate v1.36.5 adds Google AI Studio model support to the multi2vec-google module. • Adds Google AI Studio model support to the `multi2vec-google` module, expanding multimodal vectorization beyond Vertex AI.

Weaviate v1.36.0

Tue, 24 Feb 2026 01:01:19 GMT

Weaviate v1.36.0 promotes server-side batching, object TTL, backup restore cancellation, and inverted-index dropping to GA, and brings HFresh vector index into preview. • Adds wildcard support for `include`/`exclude` class lists in backup configuration, letting operators target collections by pattern. • Introduces a debug abort endpoint for the object TTL subsystem (via PR #10543) to force-stop in-flight TTL deletion cycles. • Exports new backup statuses in the OpenAPI/Swagger spec, enabling typed client integration with in-flight restore state. • Object TTL reaches GA with batch deletions, throttled pause-every-X-batches cadence, Prometheus metrics, inactive-tenant handling, RBAC `data delete` permission enforcement, and schedule-based enforcement (TTL on a collection is only allowed when a schedule is configured). • Alter Schema gains the ability to drop inverted indices from existing properties, with RBAC integration and multi-tenancy support, reclaiming disk space without recreating collections. • Adds VoyageAI V4 model support to the VoyageAI integration module. • HNSW snapshots are now enabled by default, improving restart performance without manual configuration. • Adds non-blocking segment deletions to reduce latency spikes during LSM compaction.

Weaviate v1.34.15

Fri, 20 Feb 2026 20:47:48 GMT

Weaviate v1.34.15 adds a debug endpoint for LSM bucket views, batch logic for text2vec-google, and recursive nested property resolution. • Adds a debug endpoint for holding consistent views on LSM buckets, aiding low-level storage diagnostics. • Adds naive batch logic to the `text2vec-google` module, enabling batched embedding requests. • Returns all nested object properties recursively when specified implicitly in a query.

Weaviate v1.33.17

Fri, 20 Feb 2026 10:55:07 GMT

Weaviate v1.33.17 adds batch vectorization support in text2vec-google and recursive nested property retrieval. • Adds naive batch logic to the `text2vec-google` module, enabling batch vectorization requests to Google's text embedding APIs. • Returns all nested object properties recursively when a nested object is specified implicitly in a query.

Weaviate v1.35.3

Thu, 15 Jan 2026 08:02:28 GMT

Weaviate v1.35.3 adds video modality support to the multi2vec-voyageai module and exposes backup size on status responses. • Adds video modality support to the `multi2vec-voyageai` module, enabling multimodal embeddings that include video inputs. • Returns backup size in the backup status response, giving operators visibility into backup storage consumption.

Weaviate v1.34.9

Thu, 15 Jan 2026 07:53:21 GMT

Weaviate v1.34.9 adds video modality support to the multi2vec-voyageai module. • Adds video modality support to the `multi2vec-voyageai` module, enabling multimodal vectorization of video content.

Weaviate v1.33.12

Thu, 15 Jan 2026 07:46:31 GMT

Weaviate v1.33.12 adds backup size reporting and video modality support for the multi2vec-voyageai module. • Adds video modality support to the `multi2vec-voyageai` module, enabling multimodal embeddings that include video inputs. • Returns backup size in the backup status response, giving operators visibility into backup storage usage.

Weaviate v1.35.0

Thu, 18 Dec 2025 09:45:13 GMT

Weaviate v1.35.0 adds object TTL, HFresh vector index, multi2multivec-weaviate module, zstd backup compression, and internal gRPC clustering. • Adds Object TTL (Time To Live) support: objects automatically expire based on configurable time-to-live settings, with a dedicated TTL status endpoint, a post-search filter for TTL-based queries, and automatic enabling of index timestamps for creation and update times. • Introduces the `multi2multivec-weaviate` module, enabling multi-to-multi-vector embeddings using Weaviate as the backend vectorizer. • Adds `dimensions` setting support to Cohere vectorizer and reranker modules for controlling embedding output size. • Adds `BaseURL` setting to Cohere's reranker module, enabling routing to custom or self-hosted Cohere-compatible endpoints. • Adds batch API support to the `text2vec-google` module, reducing latency and API call overhead for large ingestion workloads. • Adds naive batch processing logic across `text2vec` and `multi2vec` modules for more efficient vectorization during bulk operations. • Introduces VoyageAI v3.5 models and `voyage-3-large` support in the VoyageAI vectorizer module. • Adds `knowledge` setting support to the Contextual AI module integration. • Adds `kagome` tokenizer per-class user dictionary support for Japanese text processing. • Renames the SPFresh vector index to HFresh, with updated defaults, dedicated merge queue, metadata stored in LSM store, shared bucket architecture, and compressed centroids. • Adds `zstd` compression support for backups, reducing backup storage footprint. • Makes file chunk size configurable for the file replication service in distributed deployments. • Introduces an internal gRPC server as a REST cluster API equivalent, with connection manager, maintenance mode interceptor, and gzip compression for file copy service. • Introduces `Acks` sub-message to the `BatchStreamReply` gRPC message for improved replication acknowledgement signaling. • Adds replication scaling plan with a scale URL that includes collection and replication factor parameters.

Weaviate v1.33.10

Thu, 11 Dec 2025 08:12:58 GMT

Weaviate v1.33.10 adds maintenance mode for gRPC and BaseURL support for Cohere's reranker module. • Adds `BaseURL` setting to the Cohere reranker module, enabling use of custom or self-hosted Cohere reranker endpoints. • Adds a maintenance mode interceptor to the gRPC server, allowing the server to reject requests during maintenance windows.

Weaviate v1.32.22

Thu, 11 Dec 2025 08:06:28 GMT

Weaviate v1.32.22 adds `BaseURL` support for Cohere's reranker module and a maintenance mode interceptor for the gRPC server. • Adds `BaseURL` setting to the Cohere reranker module, allowing practitioners to point the reranker at a custom or self-hosted Cohere endpoint. • Adds a maintenance mode interceptor to the gRPC server, enabling controlled rejection of requests during maintenance windows.

Weaviate v1.34.2

Sun, 30 Nov 2025 11:18:10 GMT

Weaviate v1.34.2 adds a `/debug/config` endpoint to inspect live node configuration and makes file replication chunk size configurable. • Adds `GET /debug/config` endpoint to dump the current node configuration at runtime, with sensitive data automatically omitted. • Makes file chunk size configurable for the file replication service. • Adds support for the `knowledge` setting in the Contextual AI API module integration.

Weaviate v1.33.7

Sun, 30 Nov 2025 08:18:29 GMT

Weaviate v1.33.7 adds a `/debug/config` endpoint to inspect live node configuration without exposing sensitive data. • Adds `GET /debug/config` endpoint to dump the running node's configuration at runtime, with sensitive data automatically redacted.

Weaviate v1.32.19

Sat, 29 Nov 2025 09:30:33 GMT

Weaviate v1.32.19 adds a `/debug/config` endpoint to inspect live node configuration without exposing sensitive data. • Adds `GET /debug/config` endpoint to dump the current node configuration at runtime, with sensitive data automatically redacted.

Weaviate v1.34.1

Thu, 27 Nov 2025 18:47:10 GMT

Weaviate v1.34.1 adds VoyageAI v3.5 models, replication scaling, zstd backup compression, and new debug endpoints. • Adds a replication scaling plan with a dedicated URL endpoint accepting `collection` and `replication factor` parameters for dynamic replica management. • Adds a debug endpoint to get and set `gomemlimit` dynamically at runtime. • Adds a debug endpoint for setting max CPUs dynamically at runtime. • Adds `zstd` compression support for backups. • Adds Kagome tokenizer per-class user dictionary support. • Introduces VoyageAI v3.5 models in the VoyageAI module. • Adds `dimensions` setting support in Cohere modules. • Adds batch API support in the `text2vec-google` module. • Adds naive batch logic for `text2vec` and `multi2vec` modules. • Introduces an internal gRPC server as a REST clusterapi equivalent. • Adds retry logic to the usage GCS module during metrics upload.

Weaviate v1.33.6

Thu, 27 Nov 2025 18:46:30 GMT

Weaviate v1.33.6 adds zstd backup compression, dynamic memory/CPU debug endpoints, and an internal gRPC cluster server. • Adds debug endpoint to get and set `gomemlimit` dynamically at runtime without restarting the service. • Adds debug endpoint to set max CPUs dynamically at runtime. • Adds zstd compression for backups, reducing backup storage size. • Introduces an internal gRPC server as the REST clusterapi equivalent for inter-node communication. • Adds naive batch logic for `text2vec` and `multi2vec` modules to improve vectorization throughput. • Adds retry logic to the usage GCS module during metrics upload.

Weaviate v1.32.18

Thu, 27 Nov 2025 18:46:03 GMT

Weaviate v1.32.18 adds debug endpoints for runtime memory/CPU tuning, zstd backup compression, and naive batch logic for vectorizer modules. • Adds a debug endpoint to dynamically get and set `gomemlimit` at runtime without restarting the service. • Adds a debug endpoint to dynamically set the maximum number of CPUs at runtime. • Adds zstd compression support for backups, reducing backup size and transfer time. • Adds naive batch logic for `text2vec` and `multi2vec` modules to improve vectorization throughput. • Adds retry logic to the usage GCS module during metrics upload for improved reliability.

Weaviate v1.33.5

Mon, 24 Nov 2025 08:06:24 GMT

Weaviate v1.33.5 adds Geo HNSW index config, replication scaling API, VoyageAI v3.5 models, Cohere dimensions setting, and Google batch API support. • Adds a replication scaling API endpoint accepting `collection` and `replication factor` parameters via the updated `replication scale` URL to trigger replica scaling operations. • Adds `dimensions` setting support in Cohere vectorizer modules, enabling control over output embedding dimensions. • Adds batch API support in the `text2vec-google` module for more efficient bulk vectorization. • Adds ability to configure Geo HNSW Index settings, enabling tuning of the HNSW index for geo-type properties. • Introduces VoyageAI's v3.5 models as supported embedding options in the VoyageAI module. • Adds support for Amazon Nova Multimodal Embeddings model in Weaviate modules. • Adds support for the newest Anthropic models in the `generative-anthropic` module.

Weaviate v1.32.17

Mon, 24 Nov 2025 08:02:34 GMT

Weaviate v1.32.17 adds VoyageAI v3.5 models, Amazon Nova Multimodal Embeddings, Geo HNSW config, replication scaling, and Cohere dimensions support. • Adds `dimensions` setting to Cohere modules for controlling embedding output size. • Adds Geo HNSW Index settings configuration, enabling tunable HNSW parameters for geo-indexed collections. • Adds a replication scaling plan API, updated to include collection and replication factor parameters in the scale URL. • Adds support for batch API in the `text2vec-google` module for higher-throughput vectorization. • Introduces VoyageAI v3.5 models as supported embedding options. • Adds support for Amazon Nova Multimodal Embeddings model in the modules layer. • Adds support for the newest Anthropic models in the `generative-anthropic` module.

Weaviate v1.31.20

Fri, 21 Nov 2025 07:57:26 GMT

Weaviate v1.31.20 adds Geo HNSW index configuration, replication scaling, VoyageAI v3.5 models, Cohere dimensions, and Google batch API support. • Adds configurable HNSW index settings for Geo indexes, enabling tuning of the Geo HNSW index via collection configuration. • Introduces a replication scaling plan with a dedicated scale URL that includes `collection` and `replication factor` parameters, enabling dynamic replication factor changes. • Adds support for the `dimensions` setting in Cohere embedding modules, allowing control over output vector dimensionality. • Introduces VoyageAI's v3.5 models as supported embedding options in the VoyageAI module. • Adds batch API support in the `text2vec-google` module, enabling bulk vectorization requests to the Google embedding API. • Adds support for the newest Anthropic models in the `generative-anthropic` module.

Weaviate v1.34.0

Wed, 05 Nov 2025 21:14:32 GMT

Weaviate v1.34.0 adds SPFresh vector index, Flat Index rotational quantization, server-side dynamic batching, and Contextual AI modules. • Adds `spfresh` as a new vector index type, supporting custom configs, HNSW centroid indexing, uncompressed vectors, a disk queue, and dynamic posting-size calculation. • Adds rotational quantization (1-bit and 8-bit RQ) support to the Flat Index, with usage metrics for Flat RQ index. • Adds server-side dynamic batching (beta) to reduce client-side batching complexity. • Adds Contextual AI Generative and Reranker module integration. • Adds support for Amazon Nova Multimodal Embeddings model in the embeddings module. • Adds support for newest Anthropic models in the `generative-anthropic` module. • Adds configurable Geo HNSW Index settings via the `geo` feature. • Switches the default filter strategy to `acorn` for improved filtered vector search performance. • Introduces HNSW snapshots v3 for faster snapshot handling.

Weaviate v1.33.3

Wed, 29 Oct 2025 18:36:42 GMT

Weaviate v1.33.3 adds Multi-DC support via separate advertise and bind addresses in memberlist-raft networking. • Introduces separation between advertise and bind addresses in the memberlist-raft network layer to support Multi-DC deployments. • Allows RQ (rescoring quantization) bits to be exported when using a dynamic index.

Weaviate v1.32.15

Wed, 29 Oct 2025 08:50:37 GMT

Weaviate v1.32.15 adds Multi-DC support via separate advertise and bind addresses in memberlist-raft networking. • Introduces separation between advertise and bind addresses in memberlist-raft networking to support Multi Data Center deployments. • Allows Rescoring Quantization (RQ) bits to be exported when using a dynamic index.

Weaviate v1.33.2

Sat, 25 Oct 2025 08:23:08 GMT

Weaviate v1.33.2 adds sortable backup listings with size info and GOMEMLIMIT reporting to the usage module. • Backup list API now supports descending/ascending sort order and returns backup size in results. • Adds `GOMEMLIMIT` to the usage module payload, exposing Go memory limit data in usage reporting.

Weaviate v1.32.14

Fri, 24 Oct 2025 19:22:17 GMT

Weaviate v1.32.14 adds backup list sorting with size reporting and GOMEMLIMIT to usage module payload. • Backup list API now supports ascending/descending sort order and returns backup size per entry. • Adds `GOMEMLIMIT` to the usage module payload, surfacing Go memory limit data in usage telemetry.

Weaviate v1.33.1

Fri, 17 Oct 2025 11:25:08 GMT

Weaviate v1.33.1 adds debug endpoints for shard/lock monitoring, image support in generative-Cohere, slow-query sampling, and a broad new set of internal observability metrics. • Adds debug endpoints for shard and lock status monitoring (PR #9402). • Adds image support in the `generative-cohere` module, enabling multimodal generative queries. • Adds 1% sampled queries to the slow query log for lightweight production query tracing. • Adds new compaction metrics, async replication metrics, LSM WAL recovery metrics, memtable flushing metrics, bucket lifecycle metrics, segment metrics, LSM cursor metrics, and bucket read/write ops metrics to Weaviate's Prometheus-compatible metrics surface. • Sets RoaringSet as the default strategy for the dimensions bucket, improving dimension-tracking efficiency. • Renames the environment variable for fast failure detection to `MEMBERLIST_FAST_FAILURE_DETECTION`.

Weaviate v1.31.17

Fri, 17 Oct 2025 06:47:24 GMT

Weaviate v1.31.17 adds image support in generative-Cohere, new debug endpoints, slow-query sampling, and expanded storage/replication metrics. • Adds debug endpoints for shard and lock status monitoring (`feat(debug): add debug endpoints for shard and lock status monitoring`). • Adds 1% sampled queries to the slow query log for performance visibility. • Adds image support in the `generative-cohere` module. • Adds new compaction metrics to Prometheus instrumentation. • Adds async replication metrics. • Adds LSM WAL recovery metrics. • Adds memtable flushing metrics. • Adds bucket lifecycle metrics. • Adds segment metrics. • Adds bucket read/write ops metrics. • Sets RoaringSet as the default strategy for the dimensions bucket. • Defaults `RAFT_TIMEOUTS_MULTIPLIER` to 5 to better handle heavy-load environments.

Weaviate v1.32.11

Fri, 10 Oct 2025 07:26:25 GMT

Weaviate v1.32.11 adds image support in generative-cohere, new debug endpoints, and a broad set of new observability metrics. • Adds debug endpoints for shard and lock status monitoring via feat(debug) additions. • Adds new compaction metrics for LSM storage observability. • Adds async replication metrics for tracking replication health. • Adds LSM WAL recovery metrics. • Adds memtable flushing metrics. • Adds bucket lifecycle metrics. • Adds segment metrics. • Adds LSM cursor metrics. • Adds bucket read/write ops metrics. • Adds image support in the `generative-cohere` module. • Renames environment variable to `MEMBERLIST_FAST_FAILURE_DETECTION` for memberlist failure detection configuration. • Sets RoaringSet as the default strategy for the dimensions bucket. • Defaults `RAFT_TIMEOUTS_MULTIPLIER` to `5` to better handle heavy load environments.

Weaviate v1.33.0

Thu, 25 Sep 2025 08:48:25 GMT

Weaviate v1.33.0 adds Collection Aliases (GA), 1-bit RQ compression, ContainsNone/Not filters, and OIDC group management. • Adds `GET /alias` endpoint and alias-by-name resolution across objects, batch delete, collection GET, and shard-status operations, with RBAC requiring `read=true,collection=*` permission for alias lookup. • Supports backup and restore of aliases as part of a collection, including an `overwrite_alias` flag during restore. • Introduces `ContainsNone` and `Not` filter operators for inverted-index queries. • Adds 1-bit Rotational Quantization (`rq-1`) compression mode alongside the existing 8-bit variant (`rq-8`), with RQ bit counts now tracked in usage metrics. • Enables API-based vectorizer modules by default, removing the need for explicit opt-in configuration. • Expands vectorizer support to additional property types, and skips vectorizing objects that have only empty property values. • Adds OIDC group claim parsing from string, improving OIDC role group management support. • Switches the default filter strategy to `acorn` for improved filtered-search performance. • Adds a Weaviate health-check file for liveness/readiness probing. • Introduces experimental server-side batching capability. • Sets a default compression algorithm for vector indexes when none is explicitly configured. • Adds checksum validation to the inverted index format for data integrity.

Weaviate v1.32.8

Thu, 11 Sep 2025 16:35:54 GMT

Weaviate v1.32.8 adds `reasoningEffort` and `verbosity` params to generative-OpenAI and introduces WAND slow-log tracing. • Adds `reasoningEffort` and `verbosity` parameters to the `generative-openai` module for controlling reasoning depth and output verbosity. • Adds WAND slow-log tracing and context-cancellation support to surface slow query paths.

Weaviate v1.31.14

Wed, 10 Sep 2025 17:27:46 GMT

Weaviate v1.31.14 adds `reasoningEffort` and `verbosity` params to the generative-openai module. • Adds `reasoningEffort` and `verbosity` parameters to the `generative-openai` module for controlling OpenAI reasoning model behavior.

Weaviate v1.32.6

Fri, 05 Sep 2025 10:24:05 GMT

Weaviate v1.32.6 adds Amazon Nova model support, a new multi2vec-aws module, a text2vec-morph module, and alias-based GET collection operations. • Adds `multi2vec-aws` module, enabling multimodal vectorization via AWS. • Adds `text2vec-morph` module for text vectorization via Morph. • Adds support for Amazon Nova models in the modules layer. • Adds `maxTokens` support in the `generative-aws` module. • Supports GET collection operations via alias, expanding alias-based API coverage.

Weaviate v1.31.13

Wed, 03 Sep 2025 06:07:29 GMT

Weaviate v1.31.13 adds support for Amazon Nova models in the modules integration. • Adds support for Amazon Nova models in the Weaviate modules integration.

Weaviate v1.32.4

Tue, 19 Aug 2025 15:47:48 GMT

Weaviate v1.32.4 adds alias backup support, BlockMax AND BM25 optimization, and text2vec-google dimensions setting. • Supports a `dimensions` setting for the `text2vec-google` module, with a default of 768 for the `gemini-embedding-001` model. • Adds backup and restore support for collection aliases as part of collection backups. • Changes the default model for `text2vec-google` to `gemini-embedding-001`. • Adds a `debug` parameter to grouped generative search. • Backports BlockMax AND optimization for BM25 queries.

Weaviate v1.30.13

Thu, 24 Jul 2025 11:41:11 GMT

Weaviate v1.30.13 adds custom OIDC JWKS URL support and a new built-in read-only role. • Adds support for a custom OIDC JWKS URL, enabling custom identity provider configurations. • Adds a new built-in `read-only` role for RBAC authorization.

Weaviate v1.31.7

Wed, 23 Jul 2025 20:25:24 GMT

Weaviate v1.31.7 adds support for custom OIDC JWKS URLs. • Adds support for custom OIDC JWKS URL configuration, enabling use of non-standard OIDC providers.

Weaviate v1.31.6

Thu, 17 Jul 2025 07:40:15 GMT

Weaviate v1.31.6 adds jina-embeddings-v4 support, filtered search with MuVera, a new read-only built-in role, and AWS IAM for OIDC certificate download. • Adds support for new built-in `read-only` role for role-based access control. • Adds support for `jina-embeddings-v4` model in the Jina embeddings integration. • Adds AWS IAM authentication support when downloading OIDC certificates. • Enables filtered search with MuVera (multi-vector) indexing. • Adds ability to pass any object property to generative prompts. • Adds OIDC audit log configuration. • Enables reading of segment files with extra info. • Adds metrics for lazy segment loading.

Weaviate v1.32.0

Tue, 15 Jul 2025 00:06:14 GMT

Weaviate v1.32.0 adds collection aliases, rotational quantization, replica movement, compressed vector connections, and new embedding modules. • Adds `REPLICA_MOVEMENT_DISABLED` environment variable to control replica movement (renamed from `REPLICA_MOVEMENT_ENABLED`). • Renames `transferType` to `type` in `schema.json` for replication operations. • Renames `nodeId` to `targetNode` in the `ListReplication` API response. • Adds timestamp fields for status changes to replication operation details endpoint. • Adds Collection Alias (preview): create, update, delete, and resolve aliases for collections via new alias endpoints, usable in GraphQL schema and gRPC Search. • Adds Rotational Quantization as a new vector compression/quantization method. • Adds Compressed Vector Connections, enabling HNSW graph traversal using compressed vectors for neighbor lookups. • Adds support for reranking with the Cohere V3.5 model via the reranker-cohere module. • Adds `text2vec-google` module support for Gemini embedding models. • Renames the `text2colbert-jinaai` module to `text2multivec-jinaai`. • Adds support for the `jina-embeddings-v4` model in the JinaAI integration. • Adds `multi2multivec-jinaai` module for multimodal-to-multi-vector embeddings via JinaAI. • Adds neartext search support to the bigram module. • Adds a Cluster Usage Module for internal collection of object storage size, vector storage size, and backup file sizes in bytes cluster-wide. • Adds Cost-Aware Sort query planner with inverted-index sorter, delivering 2–200x faster filtered queries. • Adds Router with single-tenant and multi-tenant support for replica movement operations. • Improves the replica movement details endpoint with additional status information.

Weaviate v1.31.5

Fri, 04 Jul 2025 15:23:39 GMT

Weaviate v1.31.5 adds Gemini embedding support, a backups listing endpoint, and renames the JinaAI multi-vector module. • Adds `text2multivec-jinaai` module, renamed from `text2colbert-jinaai`, to reflect its multi-vector capability. • Adds support for Gemini embedding models in the `text2vec-google` module.

Weaviate v1.30.11

Fri, 04 Jul 2025 07:39:03 GMT

Weaviate v1.30.11 adds Gemini embedding model support and renames the JinaAI multi-vector module. • Adds Gemini embedding model support to the `text2vec-google` module. • Renames the `text2colbert-jinaai` module to `text2multivec-jinaai`.

Weaviate v1.31.1

Fri, 13 Jun 2025 07:53:22 GMT

Weaviate v1.31.1 adds Cohere V3.5 reranking, cost-aware query planning (2–200x faster sorts), runtime slow-log overrides, and neartext search on bigram indexes. • Adds support for the Cohere V3.5 reranking model in the reranker integration. • Adds a cost-aware query planner and inverted-index sorter, delivering 2–200x faster sorted queries. • Enables overriding query slow-log settings at runtime without a restart. • Adds neartext search support to bigram indexes. • Allows RBAC configurations with no root users defined. • Adds more information to the details endpoint for replica operations. • Improves memory performance by always reading fully loaded segments from memory and disabling bloom filters for in-memory segments.

Weaviate v1.31.0

Fri, 30 May 2025 21:46:18 GMT

Weaviate v1.31.0 adds MUVERA encoding, HNSW snapshotting, BM25 AND/OR operators, replica movement APIs, and backward-compatible named vectors. • Adds `GET /v1/replication/replicate/{id}` endpoint to query the status of a replica movement operation by UUID. • Adds `DELETE /replications/replicate` endpoint to cancel or delete replication operations. • Adds `transferType` parameter to replication API to distinguish between copy and move operations. • Adds `replicate` domain to RBAC, enabling access control over replica movement operations. • Adds `minimumOrTokensMatch` argument to BM25 keyword search, supporting AND/OR operator semantics via minimum-should-match logic. • Introduces MUVERA encoding for multi-vector representation, with configurable repetitions. • Introduces HNSW periodic snapshotting to accelerate index recovery and reduce WAL replay on restart. • Adds Prometheus metrics for FSM state transitions and replication engine lifecycle callbacks, plus a Grafana dashboard for monitoring the replication engine. • Adds a shard filter to the node/class status internal and HTTP endpoints for scoped status queries. • Enables adding new named vectors to existing collections by default, with auto-schema now producing named vectors. • Allows legacy vector to be referenced as the `default` named vector in mixed collections.

Weaviate v1.29.5

Wed, 07 May 2025 09:02:24 GMT

Weaviate v1.29.5 adds named Vectors to GroupHit responses and new metrics for OpenAI operations, shard status, and auto tenant operations. • Adds named `Vectors` to the `GroupHitAdditional` struct, exposing named vector results in group-by query responses. • Adds metrics for OpenAI operations to improve observability of OpenAI integration usage. • Adds a metric for internal shard status tracking at the DB layer, including `shard shutdown` as a valid tracked state. • Adds metrics for auto tenant activation and deactivation operations. • Introduces an optimized `mmap` package and migrates segment reads to it, reducing memory overhead for large datasets. • Improves BM25 block scoring by using a better average property length calculation for max impact scoring. • Adds a downgrade path from 1.30 to 1.29 for RAFT snapshots, enabling version rollbacks without losing RBAC state. • Sets `NoLegacyTelemetry` flag on the `raft` config to suppress legacy telemetry noise.

Weaviate v1.30.1

Wed, 16 Apr 2025 07:32:13 GMT

Weaviate v1.30.1 adds DB user last-used tracking, a BM25 block reindex REST trigger, and a configurable RAFT trailing-logs setting. • Adds a REST call to trigger BM25 block (blockmax) reindexing by initiating a shard reinit, enabling on-demand reindex without a restart. • Adds an environment variable to set a higher segment inspection limit for BM25 block searches. • Adds 'last used time' tracking to DB users, surfaced through the `/users/db` endpoint. • Returns the first 3 characters of an API key in API key response payloads, enabling key identification without exposing the full secret. • Adds configurable collections, properties, and tenants selection to the blockmax migrator, allowing targeted migration rather than full-index migration.

Weaviate v1.30.0

Thu, 03 Apr 2025 19:39:07 GMT

Weaviate v1.30.0 ships runtime config management, dynamic user/API-key REST APIs, dynamic RAG model selection, BlockMax WAND BM25, and multi-value vector GA. • Adds `maximum_allowed_collection_limit` as a runtime-configurable variable via the runtime config manager, enabling live tuning without restarts. • Adds `AUTOSCHEMA_ENABLED` as a runtime override, controllable through the runtime config manager without a restart. • Adds `ASYNC_REPLICATION_DISABLED` as a runtime override, controllable through the runtime config manager without a restart. • Adds an environment variable to enable dynamic (DB) user management (`DYNAMIC_USERS_ENABLED`, later renamed); enables REST API-driven creation, update, suspension, activation, and revocation of users and API keys at runtime. • Adds RBAC tenant filtering to batch object operations and `POST batch/references`, giving role-based access control coverage over batch workflows. • Adds RBAC filtering to the nodes endpoint so only nodes the caller has permission to see are returned. • Adds a `creationTime` field to dynamically created users and saves the first letters of the API key for identification. • Introduces the xAI generative module, adding xAI as a supported provider for retrieval-augmented generation. • Dynamic RAG model selection is now GA: select the generative model per query at runtime; supports image inputs split across `images` and `imageProperties` fields in the dynamic provider. • Adds `ENABLE_EXPERIMENTAL_DYNAMIC_RAG_SYNTAX` environment variable as a fallback option for enabling dynamic RAG syntax. • BlockMax WAND-based BM25 is now GA and enabled by default, delivering significantly faster BM25 keyword search with an online, zero-downtime migration process for existing indexes. • Multi-value vector search (ColBERT-style embeddings) is now GA; all multi-vector indexes now support BQ, PQ, and SQ quantization options. • Adds metrics support for the internal `http` server, expanding observability coverage. Breaking changes: • BlockMax WAND migration produces segment files that are not backwards compatible with previous Weaviate versions; rolling back to an earlier version after migration is not supported.

Weaviate v1.29.0

Mon, 17 Feb 2025 21:14:48 GMT

Weaviate v1.29.0 brings RBAC GA, async replication with Merkle Trees, ACORN random re-entry, and multi-vector (ColBERT) preview • Adds `EXPERIMENTAL_AUTHORIZATION_RBAC_READONLY_ROOT_GROUPS` environment variable to configure read-only root groups for RBAC. • Adds `EXPERIMENTAL_AUTHORIZATION_READONLY_GROUPS` environment variable to designate read-only groups in RBAC. • Adds group assignment/revocation endpoints for RBAC, allowing roles to be assigned to and revoked from groups (restricted to root users only). • Adds scope-based actions for role permissions in RBAC, with `MATCH` as the default scope (migrated via Raft). • Adds filter-based authorization for READ ALL operations in RBAC, covering schema, tenants, roles, and object reads. • Adds user permissions management to RBAC, enabling per-user permission assignment. • Adds RBAC permission body validation on assignment requests. • Adds separate tenant and collection controls inside the RBAC schema permission model. • RBAC moves to GA — fine-grained access control for collections, tenants, objects, and references is now production-ready. • Adds Async Replication using Merkle Trees (hashtrees) to propagate missing objects across cluster nodes efficiently. • Adds ACORN random re-entry strategy to improve vector index quality after updates and deletions, reducing query latency automatically. • Adds extra environment variables to configure ACORN filter strategy behavior. • Adds gRPC Aggregate support for search, property aggregators, and meta count queries. • Adds Multi-Vector (ColBERT) retrieval support in preview, enabling multiple vectors per document for storage and search.

Weaviate v1.28.5

Fri, 14 Feb 2025 12:57:39 GMT

Weaviate v1.28.5 adds four NVIDIA integration modules, expands RBAC with group assignment endpoints, and broadens gRPC aggregate support. • Adds `reranker-nvidia` module for reranking results via NVIDIA APIs. • Adds `generative-nvidia` module for generative (RAG) workflows via NVIDIA APIs. • Adds `text2vec-nvidia` module for text vectorization via NVIDIA APIs. • Adds `multi2vec-nvidia` module for multimodal vectorization via NVIDIA APIs. • Adds `EXPERIMENTAL_AUTHORIZATION_READONLY_GROUPS` environment variable to configure read-only RBAC root groups. • Adds `EXPERIMENTAL_AUTHORIZATION_RBAC_READONLY_ROOT_GROUPS` flag for protecting root groups from modification. • Adds RBAC group assignment and revocation endpoints, allowing roles to be assigned to and revoked from groups. • Adds `users/own-info` endpoint, replacing the former `authz/own-roles` endpoint. • Adds user read permission and user permissions management to RBAC. • Adds RBAC scope-based actions for role permissions, with `MATCH` as the default scope migrated via Raft. • Adds filter-based authorization for READ ALL operations covering schema, tenants, roles, and object retrieval. • Adds RBAC authorization to the classifications API. • Adds immutable root groups to RBAC, preventing end-users from modifying them. • Expands gRPC Aggregate to support meta count queries, property aggregators, and search. • Adds support for images in dynamic RAG syntax. • Adds `weaviate_schema_collections` metric to track collection counts. • Adds `weaviate_schema_shards` metric to track total shard count per node. • Adds HTTP server metrics to main API handlers. • Adds server metrics for main gRPC handlers. • Adds a flag to disable async replication. • Parallelises local and remote shard search to improve query throughput.

Weaviate v1.27.12

Mon, 10 Feb 2025 16:22:24 GMT

Weaviate v1.27.12 adds image support in dynamic RAG syntax and parallelizes local and remote shard search. • Adds image support in dynamic RAG syntax, enabling multimodal retrieval-augmented generation queries. • Parallelizes local and remote shard search, improving query performance across distributed deployments.

Weaviate v1.28.0

Wed, 11 Dec 2024 15:19:49 GMT

Weaviate v1.28.0 previews RBAC authorization with built-in and custom roles, collection-level isolation, and full CRUD endpoints at `/authz/roles`. • Adds RBAC CRUD endpoints (e.g., `POST /authz/roles`, returning `409` on conflict) for creating, reading, updating, and deleting roles and permissions in preview. • Supports `add-permission` and `remove-permission` operations on roles via the new authz API surface. • Adds a `read_roles` field to the schema, enabling role metadata to be returned as part of collection schema responses. • Introduces built-in roles with auto-generated permissions, alongside support for fully custom roles and permissions scoped to specific collections. • Adds a `users` domain and associated actions to the RBAC permission model, enabling user-management operations to be gated by role. • RBAC policies are persisted across all Raft nodes and reloaded on restart, ensuring cluster-wide consistency. • Adds RBAC authorization coverage to GraphQL (including batch GQL), gRPC search, REST batch delete, batch references, and object/reference endpoints. • Adds an RBAC audit log component for tracking authorization decisions and pretty-printing resource paths on errors. • Enforces collection and tenant existence validation at permission-creation time.

Weaviate v1.26.12

Wed, 11 Dec 2024 11:24:40 GMT

Weaviate v1.26.12 adds VoyageAI multimodal embeddings, Ollama batch support, and a runtime log-level API. • Supports runtime log-level configuration via the API, enabling operators to adjust verbosity without restarting the service. • Adds support for the Ollama batch endpoint, improving throughput for Ollama-backed vectorization. • Adds a new VoyageAI multimodal module, enabling image and text embeddings through VoyageAI. • Adds support for `X-Goog-*` headers in Google provider clients. • Adds environment variable overrides for Azure backup block size and concurrency settings. • Adds an option to skip waiting for self-deployed modules on startup, reducing initialization delays in custom module deployments.

Weaviate v1.25.28

Tue, 10 Dec 2024 18:51:27 GMT

Weaviate v1.25.28 adds a VoyageAI multimodal module for embedding multimodal content. • Adds a VoyageAI multimodal module, enabling multimodal embeddings via VoyageAI within Weaviate.

Weaviate v1.27.8

Tue, 10 Dec 2024 11:37:32 GMT

Weaviate v1.27.8 adds a VoyageAI multimodal module for cross-modal vector search. • Adds the VoyageAI multimodal module, enabling vectorization of mixed-modality content via VoyageAI's embedding models.

Weaviate v1.27.7

Thu, 05 Dec 2024 12:08:53 GMT

Weaviate v1.27.7 adds reindex-references API, maintenance-mode toggle, Azure env overrides, and X-Goog-* header support. • Adds environment variable overrides for Azure block size and concurrency settings. • Adds support for `X-Goog-*` headers, enabling Google-specific header passthrough. • Adds an option to skip waiting for self-deployed modules on startup. • Limits backup search scope to `BACKUP_PATH` for remote backends, reducing unintended traversal. • Adds a reindex-references feature via the debug API to rebuild reference indexes. • Enables maintenance mode to be toggled on or off via the `/debug` API.

Weaviate v1.25.27

Mon, 02 Dec 2024 20:50:14 GMT

Weaviate v1.25.27 adds environment overrides for Azure block size and concurrency, plus an option to skip waiting for self-deployed modules. • Adds environment overrides for Azure block size and concurrency settings. • Adds an option to not wait for self-deployed modules on startup. • Adds support for `X-Goog-*` headers in API requests.

Weaviate v1.27.5

Thu, 21 Nov 2024 17:56:45 GMT

Weaviate v1.27.5 adds the multi2vec-jinaai multimodal embedding module. • Adds `multi2vec-jinaai` module for multimodal vectorization using Jina AI embeddings.

Weaviate v1.26.11

Thu, 21 Nov 2024 13:07:48 GMT

Weaviate v1.26.11 adds the multi2vec-jinaai multimodal embedding module. • Adds the `multi2vec-jinaai` module, enabling multimodal vectorization via Jina AI's multi2vec models.

Weaviate v1.25.26

Thu, 21 Nov 2024 10:44:34 GMT

Weaviate v1.25.26 adds the multi2vec-jinaai multimodal embedding module. • Adds `multi2vec-jinaai` module, enabling multimodal vectorization via Jina AI's embedding models.

Weaviate v1.25.25

Wed, 13 Nov 2024 10:06:18 GMT

Weaviate v1.25.25 adds the multi2vec-cohere multimodal vectorizer module and extends the Slow Log with richer query diagnostics. • Introduces the `multi2vec-cohere` module, adding Cohere-backed multimodal vectorization support to Weaviate. • Extends the Slow Log with additional information to help determine why a query is slow.

Weaviate v1.27.3

Wed, 13 Nov 2024 08:13:11 GMT

Weaviate v1.27.3 adds multi2vec-cohere to default modules and extends the Slow Log with richer query diagnostics. • Adds `multi2vec-cohere` to the default modules list, enabling multimodal Cohere embeddings without manual module configuration. • Extends the Slow Log with additional information to help determine why a query is slow.

Weaviate v1.26.9

Fri, 08 Nov 2024 16:47:09 GMT

Weaviate v1.26.9 adds the multi2vec-cohere multimodal vectorizer module. • Introduces the `multi2vec-cohere` module for multimodal vectorization using Cohere.

Weaviate v1.27.2

Fri, 08 Nov 2024 11:36:40 GMT

Weaviate v1.27.2 adds dynamic backup locations and a new multi2vec-cohere multimodal vectorizer module. • Introduces the `multi2vec-cohere` module, enabling multimodal vectorization via Cohere's API. • Adds dynamic backup locations, allowing backup destinations to be configured at backup time rather than only at startup.

Weaviate v1.27.1

Thu, 31 Oct 2024 11:32:49 GMT

Weaviate v1.27.1 adds configurable gRPC message size, HNSW visited-list pool limit, Dynamic RAG module config, and parallel compressed vector cache prefill. • Adds an option to limit the HNSW visited list pool size, enabling memory cap on high-concurrency search workloads. • Adds support for Dynamic RAG module configuration parameters, allowing per-request generative module tuning. • Allows updating generative and reranker module configurations on existing collections without recreation. • Prefills compressed (PQ/BQ) vector caches in parallel, accelerating startup time for quantized indexes. • Performs non-blocking segment drops during compaction, reducing latency spikes on write-heavy workloads. • Improves segment cleanup to reduce storage overhead over time.

Weaviate v1.27.0

Wed, 16 Oct 2024 17:40:54 GMT

Weaviate v1.27.0 adds ACORN-based HNSW filters, backup cancellation APIs, experimental read-compute scaling, dynamic RAG via gRPC, and new embedding/generative modules. • Adds a backup cancellation endpoint and a 'list backups in progress' endpoint, with `CANCELED` status now propagated across all backup API responses and a `path` field added to list backup responses. • Adds experimental read-compute scaling via a separate `querier` component (`exp/query`) supporting vector search, property filters, and object retrieval from the LSMKV store for offloaded (FROZEN) tenants. • Supports dynamic RAG syntax through the gRPC API. • Supports sending Azure OpenAI deployment ID and resource name via request headers. • Adds support for custom number of dimensions when using Azure OpenAI. • Adds `weaviate_build_info` Prometheus metric for build observability. • Adds batch-size metrics for Prometheus observability. • Adds SIMD implementation for Bitwise Hamming distance on x86 and ARM architectures. • Adds ACORN-based minority filter improvements to HNSW for more accurate filtered vector search. • Supports multiple inputs for a single target vector in multi-target vector search. • Adds a Weaviate-hosted embeddings module. • Adds a Generative FriendliAI module. • Adds support for the JinaAI reranker API. • Enables arrays in generative searches. • Adds `gpt-4o` model support in the Generative-OpenAI module. • Renames `generative-palm` module to `generative-google`, `multi2vec-palm` to `multi2vec-google`, and `text2vec-palm` to `text2vec-google`, with AltNames support for backward compatibility. • Adds a progress indicator for schema catchup on restart. • Adds segment cleanup for LSM storage.

Weaviate v1.26.5

Fri, 27 Sep 2024 09:55:45 GMT

Weaviate v1.26.5 adds backup cancel/list endpoints, Jina V3 and VoyageAI model support, maintenance mode, and async brute-force search limit — but is flagged BROKEN and should not be used. • Adds `ASYNC_BRUTE_FORCE_SEARCH_LIMIT` environment variable to cap brute-force search results in async mode. • Adds `MAINTENANCE_NODES` environment variable to put specific nodes into maintenance mode. • Adds a backup cancel API endpoint (backported from main). • Adds support for Jina V3 embeddings, including updating the `task_type` parameter to `task` for JinaAI V3 embedding models. • Adds support for new VoyageAI embedding models with adjusted max token values. • Adds support for the `X-Databricks-User-Agent` header in Databricks integrations. • Adds support for OpenAI's `x-request-id` response header, surfacing it in errors from generative and QnA modules. • Introduces object deletion conflict resolution for distributed setups. • Introduces a limit on nested cross-reference depth in queries. • Introduces metrics for tombstone cycle start, end, and progress. • Adds a backup list API endpoint (note: subsequently disabled in this same release). Breaking changes: • This release is marked [BROKEN] / [DO NOT USE]: a bug may cause cluster data deletion in certain setups. Upgrade to v1.26.6 instead. See https://github.com/weaviate/weaviate/issues/5971 for details.

Weaviate v1.24.25

Thu, 26 Sep 2024 13:31:28 GMT

Weaviate v1.24.25 adds backup list and cancel endpoints, Jina V3 and new VoyageAI model support, and nested cross-reference depth limits. • Adds a backup cancel endpoint, allowing in-progress backups to be programmatically stopped. • Adds a backup list endpoint for querying existing backups via the API. • Adds support for Jina V3 embeddings, including the `task` parameter (replacing `task_type`) for JinaAI V3 embedding models. • Introduces new VoyageAI models as supported embedding integrations. • Introduces a limit on nested cross-reference depth in queries to bound query complexity.

Weaviate v1.25.17

Fri, 13 Sep 2024 18:43:31 GMT

Weaviate v1.25.17 adds backup list and backup cancel API endpoints. • Adds a backup list endpoint to retrieve existing backups via the API. • Adds a backup cancel endpoint to abort an in-progress backup via the API. Breaking changes: • This release contains a bug that may result in cluster data deletion in certain setups — do not use. Upgrade to v1.25.20 instead.

Weaviate v1.26.3

Thu, 29 Aug 2024 09:05:24 GMT

Weaviate v1.26.3 adds hybrid search score cutoffs, Databricks Foundation Model API support for LLM and embeddings, and a FriendliAI generative module. • Adds cutoff threshold support for hybrid search queries, enabling score-based result filtering. • Adds support for Databricks Foundation Model API as an LLM backend. • Adds support for Databricks Foundation Model API as an embedding backend. • Adds a new generative module for FriendliAI, enabling use of FriendliAI models for RAG workflows. Breaking changes: • This release is marked [BROKEN] / [DO NOT USE]: a bug may cause cluster data deletion in certain setups. Weaviate recommends upgrading directly to v1.26.6 instead.

Weaviate v1.25.13

Thu, 22 Aug 2024 07:26:44 GMT

Weaviate v1.25.13 adds Mistral text2vec module and concurrent vectorization — but is flagged broken; upgrade to v1.25.20. • Adds Mistral `text2vec` module, enabling Mistral-backed text vectorization for collections. • Adds concurrent vectorization support, allowing multiple vectors to be computed in parallel during ingestion. Breaking changes: • This release contains a bug that may cause cluster data deletion in certain setups. It is marked [DO NOT USE]; upgrade to v1.25.20 instead.

Weaviate v1.24.23

Tue, 20 Aug 2024 17:20:35 GMT

Weaviate v1.24.23 adds an experimental repair endpoint for cluster data repair operations. • Adds an experimental repair endpoint for repairing data in a Weaviate cluster.

Weaviate v1.26.1

Tue, 23 Jul 2024 16:38:25 GMT

Weaviate v1.26.1 adds JinaAI reranker API support for improved search result ranking. • Adds support for the JinaAI reranker API, enabling JinaAI-powered result reranking in search pipelines.

Weaviate v1.26.0

Tue, 23 Jul 2024 01:10:29 GMT

Weaviate v1.26.0 adds tenant offloading to S3, multi-target vector search, scalar quantization, async replication, and improved range filters. • Adds `OFFLOAD_S3_ENDPOINT` environment variable (renamed from `S3_ENDPOINT_URL`) to configure S3-compatible object storage for tenant offloading. • Adds `FROZEN` tenant status via REST and gRPC APIs, enabling inactive tenant data to be offloaded to S3-compatible object storage to reduce compute costs. • Adds `IndexRangeFilters` property config to enable a new roaring-set range index, drastically improving performance of numeric range queries at scale. • Adds a reindex endpoint to the REST API. • Adds Scalar Quantization (SQ) vector compression, mapping floating-point vector values to integers to reduce storage size while maintaining search accuracy. • Adds async (Merkle tree-based) replication to keep replicas consistent with minimal performance impact. • Adds multi-target vector search, allowing a single query to search across multiple named vectors simultaneously via GraphQL and gRPC. • Adds `generative-anthropic` as a new generative module (`Module Generative Anthropic`). • Adds dynamic generative module syntax with GraphQL and gRPC support, enabling runtime selection of generative modules. • Adds an environment variable to disable the Go profiler setup. • Adds API-based modules (including `multi2vec-palm`) enabled by default. • Enables concurrent batch vectorization requests, improving throughput for bulk ingestion. • Changes HNSW default max connections to 32 for improved index performance. • Makes offload S3 bucket auto-creation configurable. • Enables auto tenant activation/deactivation as part of the offloading workflow. • Supports concurrent tenant update operations. Breaking changes: • Tenant activity status update requests are now limited to 100 tenants per request (official client libraries batch automatically). • The `S3_ENDPOINT_URL` environment variable is renamed to `OFFLOAD_S3_ENDPOINT`. • The `UNFROZEN` tenant status is removed; use the supported active/frozen lifecycle instead.

Weaviate v1.25.8

Thu, 18 Jul 2024 20:29:20 GMT

Weaviate v1.25.8 adds opt-in Sentry error reporting and a new flag to force full-replica shard searches. • Adds a new flag to force search to query all replicas of a shard when possible, improving search completeness in replicated deployments. • Integrates Sentry error reporting (opt-in, disabled by default) with automatic reporting of vector search failures and shard initialization errors.

Weaviate v1.25.6

Fri, 28 Jun 2024 22:50:57 GMT

Weaviate v1.25.6 adds optional forced compaction for the flat index type. • Adds optional forced compaction for the flat index type, enabling manual compaction control outside of automatic scheduling.

Weaviate v1.25.0

Fri, 10 May 2024 02:33:24 GMT

Weaviate v1.25.0 adds RAFT-based schema, batch vectorization, dynamic index switching, implicit tenant creation, and new Ollama/OctoAI modules. • Adds `RAFT_GRPC_MESSAGE_MAX_SIZE` environment variable to set the maximum gRPC message size for the RAFT subsystem. • Adds an external gRPC method for getting tenant information, enabling programmatic tenant queries via gRPC. • Adds a `GET /cluster/statistics` endpoint (cluster-aware) for retrieving cluster-wide statistics. • Adds an endpoint for checking if a tenant exists. • Returns the created tenants in the response body of `POST /tenants`. • Introduces RAFT-based schema consensus, enabling concurrent schema updates across cluster nodes and eliminating schema-update bottlenecks. • Introduces batch vectorization for OpenAI, Cohere, and VoyageAI integrations, reducing rate-limiting exposure and speeding up bulk inserts. • Introduces dynamic vector index switching to automatically transition between index types for optimal performance and efficiency. • Introduces implicit tenant creation — nonexistent tenants are created on the fly when their name is included in a batch insert (auto-tenant toggling on multi-tenancy-enabled classes). • Adds `nearVector` and `nearText` as sub-search options within hybrid search queries. • Adds `groupBy` support to hybrid search and BM25F, and adds `moveTo`/`moveFrom` and similar parameters to aggregate hybrid search. • Adds target-vector cleanup for hybrid queries via gRPC. • Introduces the `text2vec-ollama` module for local embedding generation via Ollama. • Introduces the `generative-ollama` module (including Llama 3 support) for local generative AI via Ollama. • Adds OctoAI generative and `text2vec` modules for embedding and generation via OctoAI. • Adds Command R and Command R+ model support to the `generative-cohere` module. • Adds tenant activity metrics for observability of per-tenant usage. • Increases put and batch operation timeouts to 60 seconds. • Reserves `RAFT` (all casing permutations) as a protected class name, preventing naming conflicts with the consensus subsystem.

Weaviate v1.24.7

Fri, 05 Apr 2024 18:34:19 GMT

Weaviate v1.24.7 adds the VoyageAI reranker module. • Introduces the VoyageAI reranker module for result reranking pipelines.

Weaviate v1.24.2

Tue, 12 Mar 2024 18:48:25 GMT

Weaviate v1.24.2 adds generative-mistral module, gemini-pro-vision support, and multi-transformer/CLIP module capability. • Introduces `generative-mistral` module for Mistral-backed generative search. • Adds support for `VOYAGEAI_APIKEY` environment variable for VoyageAI API key configuration. • Adds support for `gemini-pro-vision` model in the generative-google module. • Adds support for multiple transformers and CLIP modules simultaneously. Breaking changes: • The `text2vec-voyageai` module's `truncate` setting type has changed from string to bool. • The VoyageAI API key environment variable is renamed from the previous name to `VOYAGEAI_APIKEY`.

Weaviate v1.24.0

Tue, 27 Feb 2024 14:12:28 GMT

Weaviate v1.24.0 adds multi-vector per class, HNSW binary quantization, Japanese/Chinese tokenizers, and high-frequency update support. • Adds binary quantization (BQ) support for the HNSW vector index, enabling vector compression into compact binary formats to drastically reduce memory footprint while maintaining search accuracy; BQ compression can be enabled via class user config updates. • Introduces multiple vectors per class (named vectors), allowing each object to carry several independent vector representations for richer, multifaceted search and ML use cases; includes gRPC Batch API support, aggregate queries with named vectors, and `VectorConfig` update support. • Adds Japanese and Chinese tokenizer support, with dictionary files bundled directly in the Docker image. • Extends HTTP backup and restore endpoints to accept custom compression configuration, and adds a restore config object. • Changes hybrid search fusion default to relative score fusion. • Supports high-frequency updates at tens of millions per day by skipping vector reindexing when vectors are unchanged and deduplicating identical objects in batch operations. • Improves the `NotEqual` filter operator for more accurate query results. • Enables setting additional log levels for more granular observability.

Weaviate v1.23.0

Mon, 18 Dec 2023 18:27:42 GMT

Weaviate v1.23.0 adds binary quantization, lazy shard loading, auto-compression PQ, Gemini/Anyscale modules, and gRPC TLS. • Adds output verbosity option to the Nodes API with a new `verbose` level that includes per-shard metadata, and a new default of `minimal` that omits it — reducing cost of cluster-wide status queries at scale. • Adds `compressed` field to `NodeShardStatus` in the Nodes API response. • Introduces `ReturnAllNonrefProperties` bool to the gRPC `PropertiesRequest` message to control property return in search results. • Adds gRPC TLS credentials support via new config options, enabling encrypted gRPC transport. • Adds metadata filter support to the gRPC search API. • Adds geo-coordinate support to the gRPC search API. • Introduces a custom `pb.Properties` message in gRPC search results for type-aware property handling. • Adds a Generative Anyscale module for LLM-backed generative search. • Adds support for Google Gemini model via a new generative module. • Adds Mixtral-8x7B-Instruct-v0.1 to available generative models. • Adds support for Google Gecko 002 and 003 embedding models. • Introduces binary quantization (BQ) and a brute-force flat index type that runs searches directly from disk, with choice between original vectors or binary-compressed vectors. • Introduces lazy shard loading: nodes now start almost instantly by loading shards in the background, with on-demand loading when a request targets a not-yet-loaded shard. • Adds Prometheus metrics for shard lazy loading and unloading. • Introduces auto-compression: Product Quantization (PQ) triggers automatically when the in-memory vector index crosses a configured threshold. • Adds resource guardrails that set memory and thread limits to prevent OOM conditions and worker-thread swapping. Breaking changes: • The Nodes API (`GET /v1/nodes`) now defaults to `minimal` verbosity, omitting per-shard metadata from the response. Callers that relied on shard-level detail must add the `verbose` verbosity parameter to restore the previous behavior.

Weaviate v1.22.5

Fri, 24 Nov 2023 15:22:53 GMT

Weaviate v1.22.5 adds text2vec-aws and generative-aws modules for Amazon-backed vectorization and generation. • Adds `text2vec-aws` module for vectorizing data using AWS-backed embedding models. • Adds `generative-aws` module for generative AI queries powered by AWS services.

Weaviate v1.22.3

Tue, 07 Nov 2023 20:16:52 GMT

Weaviate v1.22.3 adds the text2vec-jinaai module, Cohere v3 model support, and OpenAI GPT-4 128k/GPT-3.5 preview models. • Adds support for overriding Cohere's base URL via the `X-Cohere-BaseURL` HTTP header, enabling routing to custom or proxy endpoints. • Adds support for passing OpenAI base URL via HTTP header, enabling routing to custom or proxy OpenAI-compatible endpoints. • Adds the `text2vec-jinaai` module, enabling JinaAI embeddings as a vectorization source in Weaviate. • Adds support for Cohere v3 models in the Cohere integration. • Adds support for OpenAI GPT-4 128k and GPT-3.5 preview models in the OpenAI integration.

Weaviate v1.22.0

Fri, 27 Oct 2023 13:56:43 GMT

Weaviate v1.22.0 adds async indexing, nested object storage, official gRPC API, OIDC group auth, and module vectorization expansions. • Adds experimental async indexing via the `ASYNC_INDEXING=true` environment variable, decoupling vector indexing from object creation to maximize import speed. • Adds `vectorQueueSize` field to the `/schema/{className}/shards` REST API response to expose pending async index queue depth. • Adds support for `object` and `object[]` data types, enabling full nested objects to be stored directly in Weaviate, including autoschema support for dynamic nested properties. • Adds `node_mapping` parameter to backup restore operations. • Officially supports gRPC API (with proto packages split into `v0` and `v1`), including gRPC health checks and nested object transport. • Adds OIDC group authentication support. • Adds `gpt-3.5-turbo-instruct` to the available models for the `qna-openai` module. • Adds vectorization support for `text[]` properties in the `multi2vec-bind` module. • Adds vectorization support for `text[]` properties in the `multi2vec-clip` module. • Adds automatic schema repair when cluster nodes fall out of sync. • Adds memory guard rails for batch creation to prevent out-of-memory conditions under heavy load. • Improves startup time by initializing shards in parallel. • Improves shutdown speed by shutting down shards in parallel. Breaking changes: • gRPC proto files have been split into `v0` and `v1` packages; existing gRPC clients must upgrade to the latest gRPC services.

Weaviate v1.21.3

Wed, 13 Sep 2023 11:54:22 GMT

Weaviate v1.21.3 expands gRPC support with near-text/image/audio/video search, generative search, sorting, consistency, and vectorizer auth. • Adds `nearText` search to the gRPC protocol, enabling vector similarity queries over gRPC alongside the existing REST/GraphQL path. • Adds near image, audio, and video search operators to the gRPC protocol. • Adds generative search to the gRPC protocol, bringing RAG-style queries to the gRPC surface. • Adds consistency-level control to gRPC requests, matching the consistency options available over REST. • Adds vectorizer authentication support via gRPC, so module-backed vectorizers can be authorized over the gRPC path. • Adds result sorting to the gRPC protocol. • Adds Java options to the gRPC protobuf definition, improving first-class Java client support. • Supports new Google PaLM modules via the `generative-palm` integration.

Weaviate v1.21.0

Thu, 17 Aug 2023 17:10:06 GMT

Weaviate v1.21.0 adds ContainsAny/ContainsAll operators, backup compression, inactive tenants, pread LSM support, and two new vectorizer modules. • Adds `ContainsAny` and `ContainsAll` filter operators for easier filtering on array types and tokenized text fields. • Introduces the `text2vec-gpt4all` module for local GPT4All-based text vectorization. • Introduces the `multi2vec-bind` module for multi-modal vectorization via ImageBind. • Adds opt-in `pread` as an alternative to `mmap` for LSM store access, improving performance and stability on disk-bound setups. • Backup compression support: backups can now be compressed into pre-configurable chunks, reducing file operations and lowering S3/GCS storage costs. • Adds ability to deactivate tenants (experimental) so inactive tenants consume no resources, enabling denser multi-tenant deployments on the same node. • Enforces a minimum replication factor according to system-wide configuration. • Adds a configurable nested cross-reference query limit. • Adds batch queue congestion info to node status. • Adds gRPC batching support. • Adds batch support in the reranker-transformers module. • Enables creating object references without specifying `ToClass`. • Adds NEON SIMD acceleration for L2 and dot-product distance calculations on ARM, improving HNSW vector search performance.

Weaviate v1.20.0

Thu, 06 Jul 2023 17:58:45 GMT

Weaviate v1.20 adds native multi-tenancy, autocut result filtering, RelativeScore fusion, two reranker modules, and PQ GA. • Introduces native multi-tenancy with strong tenant isolation, supporting 50,000+ tenants per node and millions of tenants with billions of objects in a multi-node cluster; enable via class schema configuration. • Adds `GET /tenants` endpoint to list tenants of a multi-tenant class, plus endpoints to create and delete tenants for a specific class. • Supports full single-tenant object CRUD, batch operations, and batch reference operations, with tenant key immutability enforced. • Extends the nodes API to surface multi-tenant class information. • Adds multi-tenancy support to `GQL Get{}` and `GQL Aggregate{}` queries, including `nearObject` and `nearText` with tenant context. • Adds replication support for multi-tenant classes. • Enables Prometheus metrics for classes with multi-tenancy enabled. • Introduces `autocut` for `bm25`, `nearVector`, `nearObject`, and `nearXXX` queries to automatically cut off unrelated results. • Adds `autocut` and a `RelativeScore` fusion algorithm to hybrid search for improved result quality. • Introduces `reranker-transformers` module for post-retrieval reranking using transformer models. • Introduces `reranker-cohere` module for post-retrieval reranking using the Cohere API. • Adds status code metrics distinguishing OK, user error, and server error responses for better observability of request success and failure rates. • Product Quantization (PQ) moves to general availability, with dynamic rescoring of results and a configurable training limit.

Weaviate v1.19.7

Mon, 12 Jun 2023 11:22:23 GMT

Weaviate v1.19.7 adds PQ rescoring, new Cohere model support, and grouped metrics options. • Adds an option to group metrics via the metrics configuration, alongside a corrected Vector Add metric. • Adds Product Quantization (PQ) with rescoring support to improve approximate nearest-neighbor search accuracy. • Adds support for new Cohere model names in both the `text2vec` and `generative` Cohere modules.

Weaviate v1.19.1

Wed, 10 May 2023 19:54:39 GMT

Weaviate v1.19.1 adds Google PaLM support via new `text2vec-palm` and `generative-palm` modules. • Adds `text2vec-palm` module to enable Google PaLM-based text vectorization. • Adds `generative-palm` module to enable Google PaLM-based generative search.

Weaviate v1.19.0

Thu, 04 May 2023 15:49:38 GMT

Weaviate v1.19 adds a gRPC search API, Cohere generative module, tunable consistency, uuid prop types, and group-by queries. • Adds a minimal gRPC API (experimental) with support for a Search endpoint, enabling lower-latency programmatic access. • Adds `generative-cohere` module, enabling Retrieval-Augmented Generation with Cohere's generative models. • Adds tunable consistency to GraphQL `Get` queries, letting callers control read consistency level per search request. • Adds `uuid` and `uuid[]` property types, indexed with roaring bitmaps for efficient UUID-based filtering. • Adds group-by arbitrary property (including reference props) in GraphQL queries, returning top-k results per group. • Enriches `text` and `text[]` tokenization with new options via `IndexFilterable` and `IndexSearchable` property settings, replacing the deprecated `string` and `string[]` data types. • Migrates the `IndexInverted` property field to separate `IndexFilterable` and `IndexSearchable` fields for finer control over inverted index behavior. Breaking changes: • Downgrading from v1.19 to v1.18 is not supported after upgrading; a backup must be created before upgrading if a downgrade may be needed. • The `string` and `string[]` data types are deprecated in favor of `text` and `text[]` with explicit tokenization options.

Weaviate v1.18.4

Mon, 24 Apr 2023 15:41:25 GMT

Weaviate v1.18.4 adds Azure support across all OpenAI modules. • Adds Azure support to all OpenAI modules, enabling use of Azure-hosted OpenAI endpoints alongside existing OpenAI integrations.

Weaviate v1.18.3

Tue, 04 Apr 2023 09:31:08 GMT

Weaviate v1.18.3 adds GPT-3.5-turbo/GPT-4 support, a `properties` field for grouped generative results, and disk-space-aware shard assignment. • Adds support for GPT-3.5-turbo and GPT-4 models in the Generative OpenAI module. • Adds `properties` field for `groupedResult` in the Generative AI (OpenAI) module to limit the number of tokens sent per request. • Assigns shards and replicas to new classes based on available free disk space rather than a fixed strategy. • Allows third-party module API key headers through in CORS preflight configuration.

Weaviate v1.18.0

Tue, 07 Mar 2023 14:53:24 GMT

Weaviate v1.18.0 adds bitmap filtering, HNSW-PQ compression, BM25/Hybrid `where` filters, Cursor API, Azure backups, and full tunable replication consistency. • Adds `BACKUP_GCS_USE_AUTH` environment variable to the `backup-gcs` module to allow alternative GCP authentication forms beyond default credentials. • Adds Cursor API to scroll through every object in a class using an ID cursor, bypassing the `QUERY_MAXIMUM_RESULTS` limit at constant cost per page regardless of scale. • Adds Azure Cloud Storage as a backup destination module, joining existing `GCS` and AWS `S3` backup providers. • Extends BM25 and Hybrid Search to support `where` filters, enabling combined keyword/vector + filter queries that were not possible in v1.17. • Adds stopword support to BM25 scoring. • Extends all remaining replicated write and read endpoints with tunable consistency levels (including `PUT` and `HEAD` for objects, batch object reads, and object existence checks); changes the default consistency level from `ALL` to `QUORUM`. • Adds automatic read-repair for replication: when Weaviate detects inconsistencies between replicas it repairs them automatically, including detection of deleted objects and concurrent repairs scaled to the configured consistency level. • Introduces bitmap indexing (`RoaringSet`) for non-text properties in the LSM store, delivering up to 1,000x faster filtering; existing datasets continue working with the old index and a zero-downtime migration path is available. • Adds optional HNSW-PQ (Product Quantization) vector compression, reducing memory footprint by 25–75% while retaining HNSW recall and performance. • Reworks BM25 scoring to use the Weak-AND (WAND) algorithm with concurrent term evaluation, yielding more than 10x throughput improvement over v1.17. • Adds API key authentication (`API_KEY` auth) that can be combined with existing OIDC authentication. • Transfers backup files between S3, GCS, and Weaviate in a streaming fashion without loading file contents into memory.

Weaviate v1.17.0

Tue, 20 Dec 2022 16:51:02 GMT

Weaviate v1.17 adds leaderless replication with tunable consistency and hybrid BM25F + dense-vector search. • Introduces leaderless replication with tunable consistency, enabling high availability and horizontal read-throughput scaling across a Weaviate cluster. • Adds hybrid search combining BM25F keyword scoring and dense vector search with rank fusion, plus standalone pure BM25 and BM25F search modes. • Supports dynamically adding nodes to a running cluster after data has already been imported. • Adds TTLs to cluster-wide transactions covering schema and classification operations. • Adjusts memtable size dynamically based on workload conditions.

Weaviate v1.16.8

Fri, 16 Dec 2022 15:24:27 GMT

Weaviate v1.16.8 adds modelVersion support in text2vec-openai to enable the text-embedding-ada-002 model. • Adds `modelVersion` setting to the `text2vec-openai` module, enabling users to select the `text-embedding-ada-002` model.

Weaviate v1.16.0

Mon, 31 Oct 2022 15:53:31 GMT

Weaviate v1.16 adds distributed multi-node backups, null/length property filtering, ref2vec-centroid, Cohere and HuggingFace text2vec modules, and a cluster nodes status API. • Adds the `ref2vec-centroid` module, enabling vectorization of objects based on the centroid of their referenced objects' vectors. • Adds the `text2vec-cohere` module for Cohere-powered text vectorization, including support for the experimental `multilingual-2210-alpha` Cohere model. • Adds the `text2vec-huggingface` module with support for the HuggingFace Inference API. • Adds an API endpoint to view cluster node status, surfacing per-node health and shard information. • Adds support for OpenID scopes configuration, allowing operators to specify required scopes for OIDC authentication. • Adds a default vector distance metric setting, letting operators define the cluster-wide default metric for new classes. • Extends one-command backups (introduced in v1.15) to distributed multi-node setups; backups from v1.15 single-node setups remain backward-compatible. • Adds null-state property indexing and filtering, enabling efficient queries to find objects where a given property is set or unset — must be activated before importing data. • Adds property-length indexing and filtering, enabling efficient queries to filter objects by the length of a property value — must be activated before importing data. • Marks all shards as read-only automatically when a configurable memory threshold is reached, preventing data corruption under memory pressure. • Allows creating class schemas with self-referential (recursive) references. • Updates the OpenAI text2vec module to use the current OpenAI embeddings API.

Weaviate v1.15.4

Tue, 11 Oct 2022 08:49:59 GMT

Weaviate v1.15.4 adds support for all AWS IAM-based authorizations. • Adds support for all AWS IAM-based authorization methods.

Weaviate v1.15.0

Wed, 07 Sep 2022 15:12:20 GMT

Weaviate v1.15 adds cloud-native backups to S3/GCS, Manhattan and Hamming distance metrics, HuggingFace and SUM-Transformers modules, and new monitoring metrics. • Adds `backup-s3` Weaviate module for backing up and restoring to/from AWS S3; requires the target bucket to already exist. • Adds `backup-gcs` Weaviate module for backing up and restoring to/from Google Cloud Storage; requires the target bucket to already exist. • Supports backing up and restoring multiple classes in a single request. • Adds monitoring for backup and restore operations via Prometheus. • Enables use of `GOMEMLIMIT` environment variable (Go 1.19) to cap Weaviate memory usage — a significant operational lever for high-memory deployments. • Adds `manhattan` distance metric as a new vector index option. • Adds `hamming` distance metric as a new vector index option. • Adds `text2vec-huggingface` module for vectorization via the HuggingFace Inference API. • Adds `sum-transformers` module for summarization use cases. • New Prometheus metrics for LSM memtable vitals (current size, operation durations), concurrent read/write requests, usage dimensions on Get requests, and vector index dimensions. • Introduces a Red-Black Tree in the LSM Store to improve performance of ordered/sequential imports. • Adds thread pooling for batch requests to improve import throughput. • Significantly reduces memory footprint of HNSW index connections.

Weaviate v1.14.0

Thu, 07 Jul 2022 16:48:51 GMT

Weaviate v1.14.0 adds Prometheus monitoring, official multi-distance-metric support, and class-namespaced REST endpoints. • Adds new REST endpoints that include the class name as a namespace — e.g. object operations scoped to a specific class — eliminating ambiguity when an ID exists in multiple classes; old ID-only endpoints remain but are deprecated and will be removed in a future version. • Officially supports `cosine`, `l2-squared`, and `dot` distance metrics in the vector index, replacing the previous experimental-only status for non-cosine metrics. • Introduces `distance` as the supported similarity field in the API, replacing `certainty` (now deprecated) wherever it appears in queries. • Adds Prometheus-compatible monitoring for import metrics, HNSW operations (inserts, deletes, cleanup), LSM store segment and compaction details, startup and crash-recovery metrics, batch-delete operations, and total imported object counts. • Adds support for aggregating `date` fields in aggregate queries.

Weaviate v1.13.2

Fri, 20 May 2022 16:27:53 GMT

Weaviate v1.13.2 previews L2 distance support, with full availability planned for v1.14.0. • Adds L2 distance metric support (preview/experimental — full support coming in v1.14.0).

Weaviate v1.13.0

Tue, 03 May 2022 14:08:51 GMT

Weaviate v1.13.0 adds faceted vector search, result sorting, timestamp filtering, batch delete by filter, and DPR transformer support. • Adds `path: ["_creationTimeUnix"]` and `path: ["_lastUpdateTimeUnix"]` filter notation after optionally including `creationTimeUnix` and `lastUpdateTimeUnix` in the inverted index — enabling timestamp-based filtering for the first time. • Adds a new `/v1/batch` endpoint supporting delete-by-filter, removing all objects that match a specified filter in one operation. • Enables combining `nearVector`, `nearObject`, `nearText`, and other `near` vector searches with `Aggregate` queries for faceted vector search; requires an explicit limit or a `certainty`/`distance` threshold. • Adds sorting of search results (reads affected objects from disk; columnar-storage optimization planned for a future release). • Supports DPR (Dense Passage Retrieval) transformer models in `text2vec-transformers`, using two separate models to encode queries and passages independently.

Weaviate v1.12.0

Tue, 05 Apr 2022 13:03:08 GMT

Weaviate v1.12.0 adds configurable stopword lists, unlimited certainty-based vector search, a Shard API, and disk-pressure auto-protection. • Adds Shard API to expose individual shard status and allow marking shards as read-only via the API, blocking writes while permitting reads. • Introduces two configurable disk-pressure thresholds: a warning threshold (e.g. 80%) that logs alerts, and a critical threshold (e.g. 90%) that automatically marks all shards on the affected node as read-only. • Enables unlimited vector search by certainty, returning all results within the desired certainty range regardless of internal limits, with a configurable global maximum to prevent out-of-memory conditions. • Adds support for turning off tokenization on `string` fields so the entire field — including spaces — is indexed as a single token, preventing unwanted partial-string matches. • Introduces a fully configurable inverted-index stopword list, applicable to exact-match queries now and in anticipation of upcoming BM25 and mixed BM25/dense-vector search support.

Weaviate v1.11.0

Mon, 14 Mar 2022 14:15:54 GMT

Weaviate v1.11.0 lets you supply your OpenAI API key at query time instead of storing it server-side. • Enables passing the OpenAI API key at query time via the `text2vec-openai` module, avoiding the need to store third-party credentials on the server.

Weaviate v1.10.0

Thu, 27 Jan 2022 14:49:06 GMT

Weaviate v1.10.0 adds OpenAI embeddings, QnA reranking, HNSW EF boundaries, and a HEAD /v1/objects/{id} existence check. • Adds `HEAD /v1/objects/{id}` endpoint that returns `204` when an object exists or `404` when it does not, without loading or unmarshaling the full object from disk. • Adds `ask: { rerank: true }` to the QnA module so that multiple answer candidates are drawn from the top-n results and re-ranked by qna-specific score rather than always extracting from the single top semantic result. • Adds `dynamicEfMin` (default `100`), `dynamicEfMax` (default `500`), and `dynamicEfFactor` (default `8`) HNSW config parameters to bound and tune automatic `ef` derivation at query time. • Adds the `text2vec-openai` module, enabling OpenAI embeddings as a vectorizer for both import and query inference with a valid OpenAI API key. • Allows importing objects without a vector when vector indexing is enabled, so vectors can be added later via an update. • Allows manually overriding the vector on a class that has a vectorizer module configured, provided the replacement vector has matching dimensions and vector space.

Weaviate v1.9.0

Fri, 10 Dec 2021 16:39:20 GMT

Weaviate v1.9.0 introduces the `multi2vec-clip` module for multi-modal image+text vectorization in a single vector space. • Adds the `multi2vec-clip` module (set via `vectorizer: multi2vec-clip` and `moduleConfig.multi2vec-clip`) enabling multi-modal vectorization of `image` (`blob`) and `text`/`string` fields within a single shared vector space, with optional per-field weighting via `weights.imageFields` and `weights.textFields`. • Adds `nearImage` search alongside `nearText` search in the `multi2vec-clip` module, supporting cross-modal queries such as text search over image-only content. • Supports base64-encoded image ingestion via `blob`-typed properties when using the `multi2vec-clip` module.

Weaviate v1.8.0

Tue, 30 Nov 2021 17:51:10 GMT

Weaviate v1.8.0 adds horizontal scaling with multi-shard indices, paginated search via `offset`, and filtered vector search improvements. • Adds `offset` pagination parameter to `GET /v1/objects?limit=25&offset=75` and GraphQL `Get { Class(limit:25, offset:75) { } }` for paging through list, vector, and filter search results. • Adds `QUERY_MAXIMUM_RESULTS` environment variable to raise the default 10,000-object pagination cap (use with caution — high values can spike memory and slow the cluster). • Adds `CLUSTER_HOSTNAME` environment variable to assign a stable node hostname, required for correct shard resolution in multi-node or docker-compose deployments. • Introduces horizontal scalability with multi-shard indices, enabling Weaviate to run as a cluster across multiple nodes with configurable sharding per class (`shardingConfig` in schema). • Introduces a Flat-Search Cutoff for filtered vector search, switching automatically to a flat scan when the filtered candidate set is small enough to make HNSW traversal suboptimal. • Adds cacheable inverted-index filter segments to accelerate repeated filtered vector searches. Breaking changes: • Upgrading to v1.8.0 triggers an automatic, irreversible on-disk data migration from the single fixed-name shard layout used in v1.7.x to the new multi-shard layout; downgrading to v1.7.x afterwards requires a pre-upgrade backup. • docker-compose deployments without a stable hostname will fail after `docker-compose down` + restart because the migrated shard is pinned to the container ID hostname; set `CLUSTER_HOSTNAME=` before first starting v1.8.0 to prevent this.

Weaviate v1.7.0

Wed, 01 Sep 2021 15:22:05 GMT

Weaviate v1.7.0 adds array datatypes, a spellcheck module with auto-correct, and transformer-based NER at query time. • Adds array primitive datatypes (`string[]`, `text[]`, `int[]`, `number[]`) to the schema, enabling lists of primitives to be stored, filtered, and aggregated like scalar properties; auto-schema automatically recognizes lists of `string`/`text` and `number`/`int`. • New `text-spellcheck` module exposes a `spellCheck` field under `_additional` in GraphQL queries, returning per-term corrections (`corrected`, `original`), a `didYouMean` suggestion, `location`, and `originalText` at query time without altering results. • New `ner-transformers` module exposes a `tokens` field under `_additional` in GraphQL queries for on-the-fly named-entity extraction from object properties, with optional `properties`, `limit`, and `certainty` parameters returning `entity`, `word`, `startPosition`, `endPosition`, `certainty`, and `property` per token.

Weaviate v1.6.0

Wed, 11 Aug 2021 10:31:20 GMT

Weaviate v1.6.0 adds zero-shot classification via `"type": "zeroshot"` in `POST /v1/classficiations` • Adds `"type": "zeroshot"` to the `POST /v1/classficiations` API, enabling zero-shot classification that works with any `vectorizer` or custom vectors — no training data required; use `"classifyProperties"`, `"sourceWhere"`, and `"targetWhere"` to control which objects and labels are classified.

Weaviate v1.5.0

Tue, 13 Jul 2021 12:05:13 GMT

Weaviate v1.5.0 rewrites storage with a custom LSM-tree engine and adds Auto-Schema for schema-free imports. • Adds Auto-Schema feature: import data objects without a pre-defined schema — Weaviate infers property types on first use, is enabled by default, and defaults are configurable via environment variables. • Replaces B+Tree storage with a custom LSM-tree storage engine, delivering import speeds more than 100% faster than previous versions at scale. Breaking changes: • The entire storage mechanism has been replaced with an LSM-tree implementation: in-place upgrades from previous versions are not possible. A new Weaviate setup must be created and all data reimported. Prior backups are not compatible with v1.5.0.

Weaviate v1.4.0

Wed, 09 Jun 2021 12:50:38 GMT

Weaviate v1.4.0 adds image vectorization via `img2vec-neural`, a new `blob` datatype, `nearImage` search, per-query `ef` tuning, and full `arm64` support. • Adds `img2vec-neural` vectorizer module with `imageFields` config in `moduleConfig` to vectorize images using neural networks; `resnet50` (pytorch and keras) supported at launch, with pytorch variant supporting `amd64`, `arm64`, and CUDA. • Adds `nearImage` GraphQL search operator to vectorize a query image at search time and retrieve results by image similarity. • Adds `"skip": true` option in `vectorIndexConfig` to bypass HNSW vector indexing entirely for classes where vectorization is unnecessary (e.g. reference-only or high-duplicate classes); defaults to `false`. • Adds `ef` field to `vectorIndexConfig` (settable at schema definition and updatable post-creation) to tune HNSW recall/performance trade-off at search time; defaults to `-1` (auto). • Introduces new primitive datatype `blob` for storing arbitrary base64-encoded binary data; `blob` fields are never indexed in the inverted index, so `valueBlob` in `whereFilters` is not supported. • Adds `AVX2` hardware-accelerated dot-product calculations for `amd64` (Intel/AMD) CPUs, improving vector import and query throughput; falls back to native Go on non-AVX2 or other architectures. • Supports the entire Weaviate stack natively on `arm64` (e.g. Apple M1); components include `Weaviate Core`, `text2vec-contextionary`, `text2vec-transformers`, `qna-transformers`, and `img2vec-neural` (pytorch only); Docker images are now published as multi-architecture images requiring no configuration changes.

Weaviate v1.3.0

Fri, 23 Apr 2021 09:13:05 GMT

Weaviate v1.3.0 adds a BERT-based Q&A module with a new `ask{}` GraphQL searcher and richer transformer model metadata via `/v1/meta`. • Introduces the `qna-transformers` module, enabling BERT-style answer extraction via a new `ask{}` searcher on GraphQL `Get { ... }` queries, configured with a `"question"` (required `string`), optional `"certainty"` (`float` 0..1), and optional `"properties"` (`[]string`). • Adds a new `_additional { answer { } }` response field containing `hasAnswer` (`boolean`), `result` (nullable `string`), `certainty` (nullable `float`), `property` (nullable `string`), `startPosition` (`int`), and `endPosition` (`int`) — surfacing extracted answers directly in query results. • Supports custom Hugging Face models for Q&A via the `semitechnologies/qna-transformers:custom` base image, compatible with `transformers.AutoModelForQuestionAnswering`. • Expands the `GET /v1/meta` endpoint to include meta information about transformer models in use across all transformer-based modules.

Weaviate v1.2.0

Mon, 15 Mar 2021 16:33:41 GMT

Weaviate v1.2.0 adds out-of-the-box transformer NLP model support via the `text2vec-transformers` module with GPU-friendly microservice architecture. • Adds `ENABLE_MODULES=text2vec-transformers` environment variable to enable transformer-based vectorization (BERT, DistilBERT, RoBERTa, etc.) without custom code. • Adds `DEFAULT_VECTORIZER_MODULE=text2vec-transformers` environment variable to set transformers as the default vectorizer across all schema classes. • Adds `TRANSFORMERS_INFERENCE_API` environment variable to point Weaviate at a separately hosted inference container, enabling GPU-optimized model serving independent of Weaviate's CPU-optimized core. • Adds `poolingStrategy` class-level module config for the `text2vec-transformers` module, accepting `masked_mean` or `cls` to control how sentence vectors are derived from word vectors. • Supports `vectorizeClassName`, `vectorizePropertyName`, and `skip` module-configuration fields on classes and properties for the `text2vec-transformers` module, mirroring the existing `text2vec-contextionary` API. • Makes `ENABLE_MODULES` a required environment variable for any module usage (including `text2vec-contextionary`), enforcing explicit module declaration. • Ships pre-built Docker inference containers for popular transformer models (e.g. `semitechnologies/transformers-inference:sentence-transformers-msmarco-distilroberta-base-v2`), with support for custom Hugging Face Hub models and local PyTorch/TensorFlow models.

Weaviate v1.1.0

Wed, 10 Feb 2021 10:52:11 GMT

Weaviate v1.1.0 adds `nearObject` GraphQL search and delivers 30–50% faster cross-reference batch imports. • Adds `nearObject` search parameter to GraphQL `Get` queries, letting you find the most similar objects to a given `id` or `beacon` in a single step — no need to first retrieve the vector and run a separate `nearVector` search; supports a `certainty` threshold. • Supports combining `nearObject` with movement operations in the `text2vec-contextionary` module. • Cross-reference batch imports are now 30–50% faster on cross-reference-heavy datasets by recognising that reference updates do not change vector positions and skipping a full re-index of affected objects.

Weaviate 0.23.0

Fri, 18 Dec 2020 11:23:35 GMT

Weaviate 0.23.0 goes standalone: drops Elasticsearch and etcd for a custom vector-first storage engine with HNSW indexing. • Replaces Elasticsearch and etcd runtime dependencies with Weaviate's own vector-first storage system, making the service fully standalone. • Introduces a custom HNSW vector index implementation with full CRUD support, Write-Ahead-Commit-Log persistence, and ongoing maintenance tasks — enabling sub-50ms 20NN-vector queries on datasets of 1–100M objects. • Adds a pluggable vector index architecture (HNSW is the first supported plugin) backed by `bolt`/`bbolt` for inverted index and object storage disk operations. • Supports running with available memory smaller than total vector size via a cache-based mem/disk strategy — no requirement to keep all vectors in RAM. • Explicitly defines the behavior of the `Like` operator (wildcard semantics, modelled after Elasticsearch wildcards). • Explicitly defines multi-word query behavior for the `Equal` operator on `string` and `text` properties: words are segmented and all segments must match; `string` splits on spaces only, `text` splits on all non-alphanumeric characters. Breaking changes: • Upgrading from `0.22.x` requires a full data reimport — live upgrade is not possible because the storage mechanism has completely changed. • The `/v1/c11y/words` endpoint is removed; use `/v1/c11y/concepts` instead. • The `?meta=true` query parameter on GET requests is removed; use `?include=...` instead. • The `meta` property in object bodies is removed; use underscore fields directly (e.g. `_classification`). • The `meta` field in cross-references is removed; use the `_classification` field directly. • The `cardinality` field on properties is removed. • The `keywords` field on classes and properties is removed. • The `Like` operator now has explicitly defined wildcard semantics instead of delegating to a third-party dependency; existing queries may behave differently. • The `Equal` operator on multi-word `string` and `text` properties now has explicitly defined segmentation behavior instead of delegating to a third-party dependency.

Weaviate 0.22.20

Fri, 27 Nov 2020 14:42:19 GMT

Weaviate 0.22.20 adds kNN classification distance fields and brings standalone mode to feature parity with ES-based mode. • Adds eight new distance fields to the `_classification` underscore prop's ref meta for kNN-classified objects: `overallCount`, `winningCount`, `losingCount`, `meanWinningDistance`, `meanLosingDistance`, `closestOverallDistance`, `closestWinningDistance`, and `closestLosingDistance`. • Standalone mode reaches feature parity with the Elasticsearch-based mode, with a production-ready release (removing all ES features) targeted for v0.23.0.

Weaviate 0.22.19

Mon, 19 Oct 2020 14:35:23 GMT

Weaviate 0.22.19 adds `_certainty` underscore prop to `Get {}` queries with `explore` set. • Adds `_certainty` underscore prop to `Get {}` queries when the `explore` parameter is set, enabling certainty scores (proximity to the search query) that were previously only available on `Explore {}`.

Weaviate 0.22.16

Tue, 15 Sep 2020 15:49:09 GMT

Weaviate 0.22.16 adds full environment-variable config support, eliminating the need for a separate config file. • Adds `ORIGIN`, `CONFIGURATION_STORAGE_URL`, `CONTEXTIONARY_URL`, `ESVECTOR_URL`, `ESVECTOR_NUMBER_OF_SHARDS`, `ESVECTOR_AUTO_EXPAND_REPLICAS`, `STANDALONE_MODE`, `PERSISTENCE_DATA_PATH`, `AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED`, `AUTHENTICATION_OIDC_ENABLED`, `AUTHENTICATION_OIDC_ISSUER`, `AUTHENTICATION_OIDC_CLIENT_ID`, `AUTHENTICATION_OIDC_USERNAME_CLAIM`, `AUTHENTICATION_OIDC_GROUPS_CLAIM`, `AUTHORIZATION_ADMINLIST_ENABLED`, `AUTHORIZATION_ADMINLIST_USERS`, and `AUTHORIZATION_ADMINLIST_READONLY_USERS` environment variables, allowing full configuration of Weaviate without a separate config file. • Expands CRUD capabilities in experimental `STANDALONE_MODE=true` standalone mode, a preview of features planned for 1.0.0.

Weaviate 0.22.15

Fri, 28 Aug 2020 10:42:25 GMT

Weaviate 0.22.15 adds optional compound-word splitting in the Contextionary and multi-threaded classification. • Adds `ENABLE_COMPOUND_SPLITTING` environment variable on the Contextionary container to split otherwise-unrecognized compound words (e.g. 'thunderstormcloud' → 'thunderstorm + cloud') during vectorization; disabled by default due to up-to-100% import-time overhead, but especially valuable for compounding languages like Dutch and German. • Both `kNN` and `contextual` classification types now run multi-threaded, using one thread per available CPU core, significantly speeding up classification on larger machines.

Weaviate 0.22.13

Fri, 10 Jul 2020 13:37:33 GMT

Weaviate 0.22.13 adds `_semanticPath` GraphQL property to trace concept paths between search terms and results. • Adds `_semanticPath` underscore property to `Get{}` GraphQL queries with `explore: {}` set, returning the concept chain (e.g. `['iphone', 'apple', 'company', 'microsoft']`) between the search term and each result; maximum `limit` for requests including `_semanticPath` is `25`; requires contextionary `v0.4.14` or later.

Weaviate 0.22.12

Fri, 26 Jun 2020 14:59:17 GMT

Weaviate 0.22.12 adds `_featureProjection` to reduce vector dimensionality for 2D/3D visualization via REST and GraphQL. • Adds `_featureProjection` underscore prop to REST `GET /v1/{kinds}/?include=_featureProjection` and GraphQL `Get {}` queries, reducing object vectors to lower-dimensional representations (default 2D) for visualization. • Supports GraphQL parameters for `_featureProjection`: `dimensions` (int, default `2`), `algorithm` (string, default `tsne`), `perplexity` (int, default min(5, len(results)-1)), `learningRate` (int, default `25`), and `iterations` (int, default `100`). • Ships `t-SNE` as the first supported dimensionality-reduction algorithm under `_featureProjection`, with the underlying algorithm designed to be exchangeable in future releases.

Weaviate 0.22.11

Wed, 24 Jun 2020 08:25:58 GMT

Weaviate 0.22.11 adds `_nearestNeighbors` underscore prop to REST and GraphQL APIs for neighbor concept discovery. • Adds `_nearestNeighbors` underscore prop to the single-object REST endpoint (`GET /v1/{kind}/{id}`) and list endpoint (`GET /v1/{kinds}`) via `?include=_nearestNeighbors` query parameter, surfacing neighboring concept data alongside standard responses. • Adds `_nearestNeighbors{}` prop support to GraphQL `Get {}` queries, allowing nearest-neighbor data to be requested alongside schema-defined props.

Weaviate 0.22.8

Wed, 17 Jun 2020 13:34:37 GMT

Weaviate 0.22.8 adds `_classification` and `_interpretation` underscore props to REST and GraphQL (use 0.22.9 instead — this release has a known regression). • Adds `_classification` underscore prop to REST (`?include=_classification`) and GraphQL, exposing classification metadata for objects that were subject to a classification — previously only available via the now-deprecated `?meta=true` REST parameter, and not available in GraphQL at all. • Adds `_interpretation` underscore prop to REST (`?include=_interpretation`) and GraphQL, exposing vectorization metadata including which words were usable, their weights, and per-concept occurrence frequency from the contextionary; requires contextionary version `...-v0.4.12` or later. • Deprecates `meta?=true/false` REST query parameter in favor of explicit `?include=_classification` and `?include=_vector` underscore props; estimated removal in `0.23.0`.

Weaviate 0.22.7

Wed, 29 Apr 2020 12:52:10 GMT

Weaviate 0.22.7 rewrites contextual classification with Information Gain and tf-idf weighting, lifting accuracy from 18% to 58% on main categories. • Rewrites the contextual classification algorithm using two new user-configurable metrics — Information Gain and tf-idf — to down-weight stop words and filler words; benchmark on the 20 Newsgroups dataset shows main-category success rate rising from 18% to 58% and granular-category (20 classes) from 10% to 42%. • Adds `informationGainCutoffPercentile`, `informationGainMaximumBoost`, and `tfidfCutoffPercentile` configuration parameters to the contextual classification to let practitioners tune word-weighting and removal thresholds for their dataset.

Weaviate 0.22.6

Mon, 06 Apr 2020 13:06:51 GMT

Weaviate 0.22.6 adds reference-count filtering so you can query objects by how many linked references they have. • Enables filtering objects by the count of their references using existing compare operators (`Equal`, `LessThan`, `LessThanEqual`, `GreaterThan`, `GreaterThanEqual`) directly on a reference path in GraphQL `where` filters — supporting queries like 'find all authors who wrote at least 2 articles' or 'show all cities with no country association'.

Weaviate 0.22.5

Wed, 01 Apr 2020 08:56:35 GMT

Weaviate 0.22.5 adds navigable API root with hypertext links and cross-reference `href` fields across REST endpoints. • The `/` path now redirects (`301 Moved Permanently`) to `/v1`, and `/v1` returns a JSON list of main API categories with documentation links instead of `404 Not Found`. • Adds an `origin` config option: when set, all root and cross-reference hyperlinks are rendered as absolute URIs, enabling correct link generation behind a reverse proxy; when unset, relative links are used. • All REST endpoints that return cross-references now include a read-only `href` field alongside the existing `beacon` field, providing an HTTP hypertext reference to the respective resource.

Weaviate 0.22.4

Thu, 05 Mar 2020 16:36:10 GMT

Weaviate 0.22.4 adds contextionary language support for German, Dutch, Italian, and Czech. • Adds contextionary language support for German, Dutch, Italian, and Czech in contextionary version `xx0.13.0-v0.4.7`, with example Docker Compose files provided for each language.

Weaviate 0.22.3

Tue, 03 Mar 2020 08:49:27 GMT

Weaviate 0.22.3 exposes object vector positions via `meta=true` on both single-object and list queries. • Adds vector position data to the `meta` object returned by `GET /v1/things` and `GET /v1/actions` when the `meta=true` query parameter is set — regardless of whether the object was part of a classification. Note: each vector is ~5 KB when JSON-encoded, so use only when necessary.

Weaviate 0.22.2

Fri, 28 Feb 2020 14:43:14 GMT

Weaviate 0.22.2 adds a `phoneNumber` primitive data type with automatic international parsing and normalization. • Adds `phoneNumber` primitive data type with two user-settable sub-fields — `input` (required, type `string`) and `defaultCountry` (optional, ISO 3166-1 alpha-2 `string`) — for storing and normalizing phone numbers. • Returns seven read-only parsed sub-fields on `phoneNumber` objects: `internationalFormatted` (string), `national` (unsigned integer), `nationalFormatted` (string), `countryCode` (unsigned integer), `valid` (boolean), `input` (string), and `defaultCountry` (string). • Full `phoneNumber` type definition available in the `openapi-specs/schema.json` Swagger specification.

Weaviate 0.22.1

Tue, 04 Feb 2020 09:32:13 GMT

Weaviate 0.22.1 adds `vectorWeights` field to `Thing` and `Action` objects for per-word vector weight control. • Adds `vectorWeights` field to `Thing` and `Action` objects in `POST /v1/things` (and actions) requests — a string-to-string key-value map where keys are words and values are math expressions (using `w` for the original weight) that override contextionary-assigned weights at vector-creation time.

Weaviate 0.21.11

Thu, 16 Jan 2020 16:43:10 GMT

Weaviate 0.21.11 adds Entity Merging to deduplicate vector search results using `closest` or `merge` grouping strategies. • Adds Entity Merging capability with `closest` and `merge` grouping strategies, letting Weaviate deduplicate results by grouping objects that describe the same real-world entity based on vector distance, controlled by a `force` parameter (0.0–1.0). • The `closest` strategy surfaces only the result closest to the query per group, while the `merge` strategy preserves original field values — string fields show all original values, numerical fields show a mean, and reference fields aggregate all references from merged objects.

Weaviate 0.21.10

Thu, 16 Jan 2020 09:57:45 GMT

Weaviate 0.21.10 adds per-class and per-property vectorization control via `vectorizeClassName` and `vectorizePropertyName` schema fields. • Adds `vectorizeClassName` boolean field at the `schema/{things,actions}` class level to control whether the class name is included in vectorization (defaults to `true`). • Adds `vectorizePropertyName` boolean field at the property level in `schema/{things,actions}` to control whether property names are included in vectorization (defaults to `false`). • Relaxes contextionary-validity requirement for class names when `vectorizeClassName: false` is set, and for property names when `vectorizePropertyName: false` is set, enabling use of arbitrary identifiers without causing import failures. • Updates contextionary dependency to version `v0.4.5` to provide more precise error messages when vectorization fails due to invalid input.

Weaviate 0.21.6

Wed, 18 Dec 2019 11:23:16 GMT

Weaviate 0.21.6 adds configurable sharding, replication, and supernode threshold controls for the vector index. • Adds `vectorIndex.numberOfShards` (integer) and `vectorIndex.autoExpandReplicas` (string) config keys to control Elasticsearch shard and replica defaults per class, mirroring Elasticsearch index-module settings. • Adds `vectorIndex.supernodeThreshold` (integer) config key to override the default threshold (100 outgoing references) at which a class is treated as a supernode.

Weaviate 0.21.5

Fri, 06 Dec 2019 17:14:25 GMT

Weaviate 0.21.5 adds `sourceWhere`, `trainingSetWhere`, and `targetWhere` filters to narrow classification runs. • Adds `sourceWhere` filter to classification API to limit which unclassified objects are processed during a classification run. • Adds `trainingSetWhere` filter to classification API to restrict the training set, usable with training-set-based types such as `'type': 'knn'`. • Adds `targetWhere` filter to classification API to restrict potential label targets, usable with direct-relationship types such as `'type': 'contextual'`.

Weaviate 0.21.4

Fri, 06 Dec 2019 13:42:47 GMT

Weaviate 0.21.4 adds `contextual` classification — no training data required, targets chosen by vector distance. • Adds `type: contextual` classification to the `classification` API payload, enabling vector-distance-based classification without training data; omit the `k` field (which is `knn`-only) when using this type.

Weaviate 0.21.2

Thu, 28 Nov 2019 11:12:47 GMT

Weaviate 0.21.2 adds dedicated `/v1/.well-known/live` and `/v1/.well-known/ready` health endpoints that bypass auth. • Adds unauthenticated liveness endpoint `GET /v1/.well-known/live` returning `204 No Content` when the Weaviate instance is alive, bypassing OIDC and other auth schemes. • Adds unauthenticated readiness endpoint `GET /v1/.well-known/ready` returning `204 No Content` when the instance is ready to serve traffic, decoupled from auth-protected endpoints like `/v1/meta`.

Weaviate 0.21.1

Fri, 15 Nov 2019 09:01:45 GMT

Weaviate 0.21.1 adds wildcard string matching via the `Like` operator in `where` filters. • Adds `Like` operator to `where` filters, enabling wildcard partial-match searches on string fields using `*` glob syntax (e.g. `valueString: "Ap*e"` matches `"Apple"` and `"Apache"`).

Weaviate 0.21.0

Wed, 13 Nov 2019 13:32:45 GMT

Weaviate 0.21.0 adds RFC 7396 merge-patch support for PATCH endpoints and reintroduces batch reference adding. • Adds RFC 7396 (`application/merge-patch+json`) merge-style patching to `PATCH /v1/things/{id}` and `PATCH /v1/actions/{id}`, replacing the previous RFC 6902 patch semantics; successful merges return `204 No Content`. • Reintroduces batch-adding of references via `POST /v1/batching/references`, restoring a capability removed in 0.20.0. Breaking changes: • `PATCH /v1/things/` and `PATCH /v1/actions/` now use merge-style (RFC 7396) patch semantics instead of RFC 6902 patch semantics; clients sending RFC 6902 JSON Patch bodies will no longer work correctly.

Weaviate 0.20.4

Wed, 06 Nov 2019 14:26:36 GMT

Weaviate 0.20.4 lets you extend the contextionary with custom concepts via a new `/v1/c11y/extensions` API endpoint. • Adds a new API (see `/v1/c11y/extensions` in the swagger spec) to extend the contextionary with custom concepts — overwrite existing concept meanings or add entirely new ones; requires contextionary service version `xxxxx-v0.4.0` or later. • Introduces the `/v1/c11y/concepts/...` endpoint family as the replacement for `/v1/c11y/words/...`, with identical behavior but a cleaner path.

Weaviate 0.20.3

Fri, 11 Oct 2019 14:33:47 GMT

Weaviate 0.20.3 lets you disable vectorization and search indexing per property via `index: false` in the schema. • Adds `index: false` to property schema definitions, allowing specific properties to be excluded from both vectorization and Elasticsearch text-based indexing; properties default to indexed when `index: true` or omitted.

Weaviate 0.20.2

Thu, 10 Oct 2019 12:15:24 GMT

Weaviate 0.20.2 adds kNN-based classification via `/v1/classifications/` and a new `meta=true` option on thing retrieval. • Adds `POST /v1/classifications/` endpoint to trigger kNN-based classification of data objects using cross-referenced schema classes as training data. • Adds `?meta=true` query parameter on `GET /things/{kinds}/{id}` to expose additional fields, including classification provenance (whether a reference was set by user input or by classification).

Weaviate 0.20.0

Fri, 27 Sep 2019 09:19:30 GMT

Weaviate 0.20.0 replaces Janusgraph/Cassandra with an esvector-only backend, merges the GraphQL Meta and Aggregate APIs, and adds forced index refresh on missing cross-references. • Replaces the Janusgraph + Elasticsearch + Cassandra stack with a single vector-optimized Elasticsearch ('esvector') backend, delivering faster listing queries, lower infrastructure footprint, and fully integrated native vector search. • Configures cross-reference denormalization depth via `vector_index.denormalizationDepth` in `config.yaml` (default: `3`), controlling how many reference levels are cached in the background for efficient traversal and filtering. • Forces an Elasticsearch index refresh when a cross-referenced object is not yet visible on the index, then retries immediately — eliminating the need for client-side retry logic when adding objects with cross-references in rapid succession. • GraphQL `Meta` API is merged into the `Aggregate` API, with grouping now an optional parameter rather than always on or always off. • Distinguishes `text` properties (mapped as Elasticsearch `text`, for full-text fields) from `string` properties (mapped as Elasticsearch `keyword`, for exact values like emails and IDs), with aggregations now supported only on `string` props. Breaking changes: • The GraphQL `Meta` API is merged into the `Aggregate` API; any queries targeting the separate `Meta` API will break. • The base unit for `geoCoordinates` search distance changed from kilometer to meter; existing query values must be multiplied by 1000. • Aggregations (e.g. top-N value counts) on `text` properties are no longer supported; only `string` properties support aggregations from 0.20.0 onward.

Weaviate 0.19.2

Thu, 08 Aug 2019 11:15:58 GMT

Weaviate 0.19.2 routes GraphQL Get() queries with `explore` arguments entirely through the esvector backend for major performance gains. • GraphQL Get() queries that include an `explore: { ... }` argument are now served entirely through the esvector backend, delivering significant performance improvements for result sets larger than 20 items.

Weaviate 0.19.0

Tue, 06 Aug 2019 09:03:21 GMT

Weaviate 0.19.0 overhauls the REST API base path, GraphQL naming, and cross-reference representation with a stable-API milestone. • Adds additional fields to `GET /v1/meta` response: running Weaviate version, connected contextionary version, and the number of words in the contextionary. • Adds `group: {type: 'closest|merge', force: }` argument to GraphQL `Get -> Things/Actions -> ClassName` fields (implementation reserved for a future release; accepted without error in 0.19.0). • Adds `...on Beacon { beacon }` inline fragment support to Cross-Refs in the GraphQL `Get` API for retrieving reference URIs (accepted in schema; errors in 0.19.0 pending implementation). Breaking changes: • API base path changed from `/weaviate/v1` to `/v1` — all existing clients and integrations must update their base URL. • `GET /meta` no longer returns schema information; schema must now be retrieved via `GET /v1/schema`. • GraphQL root-level field `GetMeta` is renamed to `Meta`. • Cross-reference field `$cref` is renamed to `beacon` in all request payloads (e.g. `POST /things`, `POST /actions`, `PUT /things/`, `PUT /actions/`, `PATCH /things`, `PATCH /actions`). • Cross-references are now always represented as an array regardless of cardinality — payloads that previously sent a single object for `atMostOne` cardinality must now send an array (e.g. `{"inCountry": [{"beacon": "..."}]}`). • No clean upgrade path from 0.18.x: internal database fields were renamed, requiring a fresh 0.19.0 instance and full data re-import rather than an in-place upgrade.

Weaviate 0.18.1

Tue, 30 Jul 2019 13:36:52 GMT

Weaviate 0.18.1 lets you combine vector `explore` and structured `where` filters in a single `Get` query. • Enables combining the `explore` argument with a `where` argument inside a `Get` retrieval query, allowing vector-based ranking alongside exact string, keyword, or geo-spatial filtering in a single request.

Weaviate 0.17.0

Mon, 22 Jul 2019 14:35:55 GMT

Weaviate 0.17.0 adds vector indexing at import time and two new GQL vector-search surfaces: `{ Local { Explore }}` and explore() in `{ Local { Get }}`. • Adds new GQL field `{ Local { Explore }}` for concept (vector-based) search across Weaviate's vector search backend. • Extends `{ Local { Get }}` with a new explore() argument enabling vector-based concept search alongside the existing where() structured search. • Imports through non-batch paths now generate and store a vector representation in Weaviate's vector database at import time. Breaking changes: • Vector-based concept search requires a new `esvector` backend — see `docker-compose/runtime/docker-compose.yml` for a reference configuration. • Vectors are only created at import time with no reindex capability, so upgrading to 0.17.x requires setting up a fresh Weaviate installation and reimporting all concepts.

Weaviate 0.16.0

Tue, 09 Jul 2019 14:32:48 GMT

Weaviate 0.16.0 lets users supply their own UUIDs on POST and batch endpoints, simplifying cross-reference imports. • Adds user-specified UUID support: set the `id` field on `POST /things`, `POST /actions`, `POST /batching/things`, and `POST /batching/actions` to assign your own UUIDs instead of letting Weaviate generate them, making cross-reference preparation possible without round-tripping for server-assigned IDs.

Weaviate 0.15.0

Mon, 24 Jun 2019 13:47:45 GMT

Weaviate 0.15.0 adds read-only user support to the AdminList authorization plugin. • Extends the `AdminList` authorization plugin with a `read_only_users` list, enabling three permission tiers: Admins (full CRUD), read-only users, and authenticated-but-unauthorized users.

Weaviate 0.14.5

Wed, 05 Jun 2019 15:28:45 GMT

Weaviate 0.14.5 adds OIDC discovery redirect at `GET /weaviate/v1/.well-known/openid-configuration` • Adds `GET /weaviate/v1/.well-known/openid-configuration` endpoint that redirects (`302 Found`) to the configured token issuer's OIDC discovery page when OIDC auth is enabled, or returns `404 Not Found` when no OIDC issuer is configured.

Weaviate 0.13.0

Tue, 21 May 2019 13:55:58 GMT

Weaviate 0.13.0: initial stable release with GraphQL traversal, pluggable backends, and OIDC auth support • Supports pluggable database backends ('connectors'), defaulting to `janusgraph` with `cassandra` for storage and `elasticsearch` as the indexing backend • Pluggable authentication and authorization providers, defaulting to `anonymous_access` with optional OIDC (Open ID Connect) configuration • Full REST API for CRUD operations on schema and concepts (Things and Actions) • Dynamic graph traversal via GraphQL, including context-based search through 'Fetch' GQL APIs • Optional asynchronous analytics jobs via Spark integration • Horizontal scaling (HA) support and 12-factor compatible configuration management • Production-quality Helm charts available (separate release lifecycle) and Docker Compose 'Try Out' setups • Ships as Docker image `semitechnologies/weaviate:0.13.0`