milvus-io/milvus

What is it?

Key features

• 01Vector search with HNSW, IVF, FLAT, SCANN and DiskANN index types
• 02Metadata filtering, range search and hybrid search with scalar fields
• 03Full-text search using BM25 and sparse embeddings such as SPLADE
• 04Distributed, Kubernetes-native deployment with independently scalable nodes
• 05Multi-tenancy isolation with RBAC, user authentication and TLS encryption
• 06Milvus Lite for local use via pip install pymilvus[milvus-lite]

Use cases

• Building retrieval-augmented generation (RAG) pipelines over private documents
• Image, text or multi-modal similarity search in production applications
• Recommendation systems that match users and items by embedding similarity
• Prototyping vector search locally with Milvus Lite before deploying a cluster

Quick start

1. 1

   Install the Python SDK pymilvus

   pip install -U pymilvus
2. 2

   Import MilvusClient from pymilvus

3. 3

   Create a local Milvus Lite client with a file name for persisting data

4. 4

   Create a collection with the client

Things to know

• !Running a self-hosted distributed Milvus cluster is complex, with many microservices on Kubernetes.
• !The lightweight Milvus Lite option is meant for quickstart use, not for production-scale workloads.

Topics

FAQ

What is milvus?

Milvus is an open-source, distributed vector database for storing and searching embeddings at scale, used for AI search and RAG applications. Milvus is a vector database that stores embeddings produced from unstructured data such as text, images and multi-modal content, and runs approximate nearest neighbor search over them. It can store vectors alongside scalar fields like integers, strings and JSON, and supports metadata filtering, range search and hybrid search.

Is milvus free to use? What license does it have?

milvus is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.

Is milvus still maintained?

Actively maintained. The last commit to milvus was on Oct 21, 2026.

What can I use milvus for?

Building retrieval-augmented generation (RAG) pipelines over private documents Image, text or multi-modal similarity search in production applications Recommendation systems that match users and items by embedding similarity Prototyping vector search locally with Milvus Lite before deploying a cluster

How do I install milvus?

Install the Python SDK pymilvus: pip install -U pymilvus Import MilvusClient from pymilvus Create a local Milvus Lite client with a file name for persisting data Create a collection with the client

Where can I get milvus?

The source code and installation instructions are on GitHub: https://github.com/milvus-io/milvus