Milvus vs pgvector
Two of the top projects in Vector databases, side by side: who leads on each score part and who each one suits, then setup, license, activity and what each review found.
Milvus
Distributed vector database with dense, sparse and hybrid search at scale
pgvector
PostgreSQL extension for vector similarity search with HNSW and IVFFlat
Verdict
Milvus ranks higher, 89 to 68. Of the 4 score parts we measure for both, Milvus leads on 3, pgvector on none and 1 is even.
- AdoptionMilvus leads by 22 points, 90 to 68: a wide gap.
- FreshnessEven at 100.
- MaintenanceMilvus leads by 23 points, 83 to 60: a wide gap.
- Agent-readyMilvus leads by 70 points, 70 to 0: a wide gap.
Pick Milvus if you are
Teams needing billion-scale vector search on Kubernetes
- Index types HNSW, IVF, FLAT, SCANN, DiskANN plus GPU CAGRA
- Dense, sparse (BM25, SPLADE, BGE-M3) and hybrid search in one collection
Pick pgvector if you are
Teams already on Postgres who want vectors without a new database
- Vectors live next to relational data with ACID, joins and point-in-time recovery
- HNSW and IVFFlat indexes with six distance operators
| What we compare | Milvus | pgvector |
|---|---|---|
| Score parts, out of 100 | ||
| Adoption | 90, widely used | 68, popular |
| Freshness | 100, active | 100, active |
| Maintenance | 83, healthy | 60, fair |
| Agent-ready | 70, partly | 0, none |
| Facts from GitHub and the README | ||
| Stars | 46.6k | 23.6k |
| License | Apache-2.0 (permissive) | custom license (read the license) |
| Last commit | Nov 2026 | Oct 2026 |
| Last release | Oct 2026 | None published |
| Language | Go | C |
| Docker | Yes | Yes |
| GPU | Optional | Not needed |
| arm64 or Apple Silicon | Mentioned | Mentioned |
Milvus
Milvus is a distributed vector database (Go and C++, LF AI & Data Foundation) that separates compute and storage on Kubernetes, with a Standalone Docker mode and pip-installable Milvus Lite. It offers HNSW, IVF, FLAT, SCANN and DiskANN indexes, GPU CAGRA, sparse BM25 and learned-sparse vectors for hybrid search, metadata filtering, multi-tenancy, hot/cold storage, auth, TLS and RBAC.
Who it is for: teams needing billion-scale vector search on Kubernetes
Strengths
- Index types HNSW, IVF, FLAT, SCANN, DiskANN plus GPU CAGRA
- Dense, sparse (BM25, SPLADE, BGE-M3) and hybrid search in one collection
- Multi-tenancy at database, collection, partition or partition-key level
- Mandatory auth, TLS and RBAC; Milvus Lite via pip for local dev
Weaknesses
- Distributed mode is Kubernetes-native with several microservices to operate
- No port, RAM or Docker command in the README; install lives in docs
- Zilliz is the major contributor and promotes its managed cloud
- Source build needs Go 1.21+, CMake, GCC 11+ and Python 3.8 to 3.11
- GPU optional
- Compose
- Compose runs Milvus, MinIO
- Models: any embedding model or service; pymilvus[model] wraps embedding and reranking models
pgvector
pgvector is a PostgreSQL extension (Postgres 13+) that adds vector, halfvec, bit and sparsevec column types with L2, inner product, cosine, L1, Hamming and Jaccard distance operators, exact search by default and HNSW or IVFFlat indexes for approximate search. Vectors sit beside ordinary rows with ACID, joins and backups, and Postgres full-text search can be combined for hybrid retrieval. It installs via make, Docker or OS packages.
Who it is for: teams already on Postgres who want vectors without a new database
Strengths
- Vectors live next to relational data with ACID, joins and point-in-time recovery
- HNSW and IVFFlat indexes with six distance operators
- Half-precision, binary and sparse vector types plus binary quantization
- Installs via make, Docker, Homebrew, APT, Yum; preinstalled on many hosted Postgres
Weaknesses
- vector type capped at 2,000 dimensions (halfvec 4,000)
- Approximate indexes filter after scanning; filtered recall needs iterative scan tuning
- HNSW builds slow down sharply once the graph exceeds maintenance_work_mem
- No server of its own; capacity depends on your Postgres tuning
- no GPU
- Docker
- Needs PostgreSQL 13+
- Models: any embedding model; stores precomputed vectors
More in Vector databases
Building with these?
We help teams pick an open-source AI stack and run it in production. Client work never changes a rank.