What it does This is a curated awesome-list that catalogs the sprawling ecosystem of vector search: standalone engines (Milvus, Qdrant, Weaviate, etc.), client libraries (Faiss, ScaNN, Annoy), hosted cloud services (Pinecone, Zilliz), and a handful of foundational research papers. Think of it as a map for the territory that every RAG app eventually has to navigate.

The interesting bit The list reveals how quickly “just use Postgres” became a serious answer—pgvector and pgANN sit right alongside purpose-built databases. It also captures the awkward adolescence of the field: some projects are full databases, some are plugins, some are browser-based experiments, and at least one claims to be a “Hippocampus For AI.”

Key highlights

• 20+ standalone services, from heavyweights (Cassandra 5.0, Solr 9.0) to newcomers (VQLite, Vexvault)
• 30+ libraries spanning C++, Java, Python, Rust, and even WASM (Voy)
• Cloud pricing models range from serverless (Pinecone, DataStax Astra) to ClickHouse-based (MyScale)
• Research section covers billion-scale ANN methods including SPANN (NeurIPS 2021) and HNSW graphs
• CC0 licensed, so fork and maintain your own fork when this one inevitably drifts out of date

Caveats

• Curation quality is uneven: some entries get detailed descriptions, others are just links with taglines copied from their own marketing
• No comparison matrix, benchmarks, or guidance on when to choose what
• “Research papers” section is thin—seven papers versus dozens of tools—and leans heavily on a few NeurIPS/ECCV hits

Verdict Worth bookmarking if you’re in the “evaluate three options before building” phase. Skip it if you already know your stack; this won’t tell you anything Faiss’s own docs or a Pinecone free tier won’t.