Milvus
What it is
Milvus is an open-source vector database developed by Zilliz for GenAI applications. Its search options include approximate nearest neighbor, filtering, range, hybrid and full-text search, with reranking, fetch and query operations. Milvus can run as Lite, a Python library; Standalone, a single-machine server; or Distributed on Kubernetes. The documentation recommends Lite for up to a few million vectors, Standalone for up to 100 million, and Distributed for 100 million to tens of billions. Lite can run locally without a cloud service, including for searches involving private or sensitive data. Official APIs and SDKs cover REST, Python, Go, Java, Node.js and C++; C# is contributed by Microsoft and Rust is under development. Security features include user authentication, TLS for gRPC and REST traffic, and role-based access controls at collection, database and instance levels. An online upgrade from Standalone to a cluster is not currently supported. Milvus is Apache 2.0 licensed and associated with the LF AI & Data Foundation under the Linux Foundation; Zilliz Cloud is linked as its managed offering.
For and against
For
- Supports hybrid, full-text and approximate nearest neighbor search
- Three deployment options for different dataset scales
- Lite can run locally without a cloud service
- Documents TLS, authentication and role-based access control
Against
- Online upgrade from Standalone to a cluster is not supported
- Rust SDK is under development
- Default 3.x deployment uses Woodpecker, etcd and MinIO
The verdict
Milvus offers multiple deployment models and a broad set of search operations, with documented scale guidance for each model. Consider the standalone-to-cluster upgrade limitation when planning a deployment path.
Who it suits
Milvus suits developers and teams building vector search into GenAI applications, with deployment options ranging from local use to Kubernetes clusters. The documentation describes Distributed as suited to enterprise users running large-scale production search systems.
Plans and prices
Milvus (open-source software)
Apache 2.0 licensed · Milvus Lite: recommended for datasets up to a few million vectors · Standalone: suitable for datasets up to 100 million vectors
Read 4 Oct 2026 · milvus.io
On the record
- Access control
- RBAC can control operations at collection, database, and instance levels.milvus.io · read 4 Oct 2026
- Dependencies
- A default Milvus 3.x deployment uses Woodpecker for messaging, etcd for metadata, and MinIO for object storage.milvus.io · read 4 Oct 2026
- Deployment
- Milvus offers Lite as a Python library, Standalone as a single-machine server, and Distributed for Kubernetes clusters.milvus.io · read 4 Oct 2026
- Deployment caveat
- The documentation says an online upgrade from a standalone Milvus instance to a cluster is not currently supported.milvus.io · read 4 Oct 2026
- Deployment limits
- The docs recommend Lite for up to a few million vectors, Standalone up to 100 million, and Distributed for 100 million to tens of billions.milvus.io · read 4 Oct 2026
- Deployment options
- Milvus offers Lite as a Python library, Standalone as a single-machine server, and Distributed for Kubernetes clusters.milvus.io · read 4 Oct 2026
- Integrations
- The integrations documentation lists tools including LangChain, LlamaIndex, Haystack, DSPy, Dify, Langflow, OpenAI, Cohere, and HuggingFace.milvus.io · read 4 Oct 2026
- Intended users
- The documentation describes Distributed as suited to enterprise users running large-scale vector search systems in production.milvus.io · read 4 Oct 2026
- Local use
- Milvus Lite can run on a device without a cloud service for local search of private or sensitive data.milvus.io · read 4 Oct 2026
- Maker
- The documentation says Milvus was developed by Zilliz and is associated with the LF AI & Data Foundation under the Linux Foundation.milvus.io · read 4 Oct 2026
- Managed option
- Milvus links to Zilliz Cloud as its fully managed offering and describes a free option on its documentation site.milvus.io · read 4 Oct 2026
- Product
- Milvus is an open-source vector database built for GenAI applications.milvus.io · read 4 Oct 2026
- Scale
- Milvus supports deployments from local notebook demos to Kubernetes clusters handling tens of billions of vectors.milvus.io · read 4 Oct 2026
- Scale limits by deployment
- The documentation recommends Milvus Lite for up to a few million vectors, Standalone for up to 100 million, and Distributed for 100 million to tens of billions.milvus.io · read 4 Oct 2026
- SDKs
- Milvus provides official REST, Python, Go, Java, Node.js, and C++ APIs or SDKs; C# is contributed by Microsoft and Rust is under development.milvus.io · read 4 Oct 2026
- Search
- Milvus supports approximate nearest neighbor, filtering, range, hybrid, and full-text search, plus reranking, fetch, and query operations.milvus.io · read 4 Oct 2026
- Search features
- Milvus supports approximate nearest neighbor, filtered, range, hybrid, and full-text search, as well as reranking and queries.milvus.io · read 4 Oct 2026
- Security
- Milvus documents user authentication, TLS encryption for gRPC and REST traffic, and role-based access control.milvus.io · read 4 Oct 2026
