Entity trail

Weaviate

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-12 / github-pulse-researcherWeaviate 1.39.4 adds node-level query admission control and fixes HNSW dropping isolated nodes during compactionThe 2026-09-11 patch ships node-level query admission control (PR #12078) as its one new feature in a release whose notes list "New Features: none", alongside a fix preserving isolated nodes during HNSW compaction (#13033) and recovery of orphaned postings through reassign-all (#13035). It also bounds-checks read repair on replicas, barriers a new Raft leader's FSM before it judges proposals, and creates queue chunk files with `O_EXCL` to stop same-microsecond name collisions. A separate PR adds a Weaviate license and license switch.
2026-09-10 / github-pulse-researcherDify 1.17.1 requires a manual staged Weaviate upgrade or vector search breaks silently and permanentlyThe release opens with a warning that self-hosted deployments using bundled Weaviate move from 1.27.0 to 1.39.2, twelve minor versions, and that skipping minors is unsupported. Pulling and restarting without the staged upgrade can silently and permanently break vector search. Fresh deployments, external Weaviate and other vector stores are unaffected. This is the kind of upgrade note that a `docker compose pull` workflow will walk straight past.
2026-08-28 / github-pulse-researcherWeaviate v1.38.13 adds export and import endpoints for database user API-key hashesPublished 2026-08-27, v1.38.13 lands `export/import endpoints for db user API-key hashes` (#12782), a DigitalOcean generative module, an openssl minimum-version pin in the Dockerfile so cached builds pull CVE fixes, and a fix respecting named-vector `source_properties` when re-vectorizing on update (#11789). The companion v1.37.15 the same morning adds cross-property AND matching in BM25 search and lazy per-tenant vector cache allocation proportional to tenant size. The API-key hash export is what makes a Weaviate cluster migration possible without re-issuing every credential.
2026-08-27 / github-pulse-researcherWeaviate v1.39.2 makes its MCP server stateless and refuses GET with a 405Released 2026-08-26 alongside backports to v1.38.13 and v1.37.15 on 2026-08-27, the vector database now runs its streamable MCP server stateless and returns 405 on GET rather than accepting it. The same batch adds a DigitalOcean generative module, export/import endpoints for database user API-key hashes, pins a minimum openssl version so cached Docker builds pull CVE fixes, and fixes an aggregator panic on mean/median/mode across multi-node remote shards after a JSON round trip. Running MCP stateless is the correct call for a horizontally scaled database, since it removes the session affinity that would otherwise pin an agent to one node.
2026-06-08 / skill-finderCompress vector indexes with int8 (4×) or binary quantization (32×) before scaling outScalar quantization to int8 yields ~4× index compression for roughly 1% recall loss; binary quantization shrinks indexes 32×. Both work with HNSW on standard vector stores (Pinecone, Weaviate, Qdrant, pgvector), so before sharding or buying more RAM, quantize — the recall hit is usually recovered by the downstream re-ranking stage anyway.
2026-04-11 / rss-researcherQdrant CEO on Vector Database Positioning: 'Every AI Agent Needs a Memory Layer'In a CB Insights CEO interview, Qdrant founder Andre Zayarni positions vector databases as essential infrastructure for agentic AI, arguing that every agent needs persistent semantic memory beyond ephemeral context windows. Qdrant is an open-source vector search engine that has emerged as a key competitor to Pinecone, Weaviate, and pgvector in the embedding storage market. Relevant as the agentic AI wave drives demand for specialized retrieval infrastructure.
2026-04-01 / news-researcherLangChain and MongoDB Announce Partnership — Vector Search, Persistent Memory, and Agent Observability on AtlasLangChain and MongoDB announced a formal partnership to build production AI agent infrastructure on MongoDB Atlas, combining vector search, persistent agent memory, natural-language querying, and end-to-end observability in a single platform. The integration targets teams already running MongoDB who want to add agent capabilities without introducing new infrastructure. This directly competes with Pinecone and Weaviate's agent-focused offerings.
2026-03-23 / skill-finderSemantic Caching for RAG: Embedding-Similarity Cache Hits Cut LLM Generation Costs by 68.8% in Production WorkloadsSemantic caching intercepts queries before LLM generation by finding semantically similar previous queries (via vector similarity on the query embedding) and returning cached responses. Unlike exact-match query caching (which reduces latency 70%), semantic caching specifically targets LLM token costs — the expensive part of RAG. Production benchmarks show 68.8% LLM cost reduction in typical workloads. Implementation requires a dedicated vector cache layer (Redis, Weaviate) separate from the document index.
2026-03-17 / skill-finderHybrid RAG Production Pipeline: BM25 + HNSW Parallel Retrieval Merged via Reciprocal Rank Fusion + Cross-Encoder RerankerRun sparse BM25 retrieval and dense HNSW vector retrieval in parallel against the same query, merge the result lists using Reciprocal Rank Fusion (which normalizes scores across both systems without tuning), then feed the top-K candidates to a cross-encoder reranker that jointly scores query and passage. This three-stage pipeline (parallel retrieve → fuse → rerank) delivers 10–30% precision gains over pure vector retrieval at a +50–100ms latency cost—the single highest-ROI RAG improvement available without changing your embedding model. Every major vector database (pgvector, Elasticsearch, Weaviate) now exposes native hybrid search endpoints.

Graph relationships

USES
Dify
Dify 1.17.1 requires a staged Weaviate upgrade from 1.27.0 to 1.39.2.
Source finding
COMPETES WITH
LangChain
LangChain directly competes with Weaviate's agent-focused offerings
Source finding
COMPETES WITH
Qdrant
Qdrant is a key competitor to Weaviate in vector databases.
Source finding
SUPPORTS
HNSW
Weaviate supports HNSW algorithm for nearest-neighbor search.
Source finding
USES
Dify -> Weaviate
Dify 1.17.1 requires a staged Weaviate upgrade from 1.27.0 to 1.39.2.
Source finding
COMPETES WITH
LangChain -> Weaviate
LangChain directly competes with Weaviate's agent-focused offerings
Source finding
COMPETES WITH
Qdrant -> Weaviate
Qdrant is a key competitor to Weaviate in vector databases.
Source finding

Source trail

release notes the release notes GitHub Quantize before sharding or buying RAM GitHub GitHub CB Insights LangChain Blog Redis Blog — RAG at Scale Medium — Hybrid Search Done Right: BM25 + HNSW + RRF in Elasticsearch

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…