Vector-database infrastructure

15 systems in the vector-database infrastructure category of the AI Agent Infrastructure Landscape, grouped by maturity tier.

Tier 1 — battle-tested (13)

  • Chroma Embedded vector DB (Rust core)

    Limited native hybrid (users build RRF custom). Fast Rust core (v2.5).

  • Databricks Vector Search Storage-optimised + Unity Catalog governance

    Storage-optimised index scaling to billions of vectors. Hosted MCP servers for UC functions, Genie, Vector Search. Agent framework integrates retrieval with Unity Catalog governance.

  • Elasticsearch / OpenSearch Mature hybrid search

    Mature hybrid (BM25 + vector) implementations. RRF + weighted combination. Elastic ELSER sparse model.

  • LanceDB Embedded Arrow-columnar vector DB

    Embedded vector DB (Arrow columnar). RRF reranker. Petabyte-scale on disk.

  • Milvus Multi-vector + hybrid + GPU

    Multi-vector columns (10 simultaneous). Native hybrid search (v2.5). CAGRA + Vamana GPU/CPU (v2.6).

  • MongoDB Atlas Vector Search Vector embedded in document DB

    Agent memory store for both short-term (document) and long-term (vector). LangGraph checkpointer for stateful agents. Vector search extended to Community Edition (Sept 2025).

  • pgvector PostgreSQL vector extension

    Stores embeddings alongside relational + full-text data. HNSW + IVFFlat ANN indexes. Used as agent conversation memory via LangChain + MCP. Foundation of Supabase AI and many self-hosted RAG stacks.

  • Pinecone Managed vector DB + cascading retrieval

    Managed vector DB. Cascading sparse + dense + rerank pipeline; pinecone-rerank-v0 .

  • Qdrant Vector DB + DBSF + RRF

    Distribution-Based Score Fusion + RRF. Sparse vectors native; filtering via ANN graph modification.

  • Snowflake Cortex Search Hybrid vector + keyword + semantic rerank

    Hybrid vector + keyword + semantic reranking. Cortex Agents (GA Nov 2025) orchestrate over Cortex Search for unstructured data. No separate vector infra needed inside Snowflake.

  • Turbopuffer Object-storage-first vector DB

    Object-storage-first vector DB. Powers Notion and Cursor. SPFresh indexing; ~10× cost reduction vs. memory-resident peers.

  • Vespa Unified vector + lexical + structured + ranking

    Combines vector ANN, lexical, structured filtering, and ML-learned ranking in one distributed system. Integrated chunking + layered ranking for RAG. No separate embedding infra needed.

  • Weaviate Hybrid vector DB + learned ranking

    Hybrid Search 2.0 with learned ranking. BM25 + vector in unified index.

Tier 2 — production-ready (2)

  • Activeloop Deep Lake Multimodal vector + serverless Postgres

    Deep Memory feature optimises embedding space per use-case (+22% retrieval accuracy). Deep Lake PG unifies serverless Postgres (agent short-term state) + billion-scale vector search (long-term memory).

  • Marqo Multimodal vector via single API

    End-to-end multimodal (text + image) vector search via single API. Handles embedding generation internally. Ecommerce personalization with proprietary models.