Vector Databases

Embedded, developer-first vector database

Best DX for prototyping and local development — in-process or client/server mode. Not recommended for high-concurrency production loads. Use to validate ideas quickly.

Vector similarity search built into the Elasticsearch engine

If you already run Elasticsearch, adding vector search avoids a new system. HNSW indexing with hybrid BM25+kNN. Mature ops tooling. Best for teams with existing Elastic infrastructure.

Embedded serverless vector database backed by Lance columnar format

Zero-dependency embedded DB that stores vectors in S3/GCS directly. Columnar Lance format makes it fast for analytics workloads. Great for edge/serverless — watch for production-scale stories.

Cloud-native distributed vector database for billion-scale

Go-to for billion-scale vector workloads. Kubernetes-native, sharding built in. Operational complexity is high — consider Zilliz Cloud (managed Milvus) for teams without K8s expertise.

Native vector search within MongoDB Atlas — no separate database needed

Store vectors alongside your application data in MongoDB. Great for teams already on Atlas — avoids a second database. Lucene-based kNN with pre-filtering on document fields.

Fully managed serverless vector database

The operator-friendly choice: zero infra, auto-scaling, metadata filtering, and namespacing for multi-tenancy. Higher cost vs self-hosted but operational simplicity wins at scale.

High-performance vector similarity search engine in Rust

Fastest single-node performance benchmark in the Open Vector Benchmark. Payload filtering is rich and efficient. Best self-hosted choice for performance-critical workloads.

Redis with vector similarity search and full-text index

Excellent for semantic cache + vector search in one service. Sub-millisecond latency. Use when you need vectors AND a cache/pub-sub layer — avoids adding another system.

Open-source vector database with built-in ML modules

Best for hybrid (vector + BM25) search out of the box. Vectorizers baked in (OpenAI, Cohere). GraphQL API is expressive. Self-host on K8s or use Weaviate Cloud.

PostgreSQL extension for vector similarity search

If you already run Postgres, pgvector eliminates an entire service. IVFFlat and HNSW indexes handle millions of vectors. Use with Supabase or neon for serverless deployment.