milvus-io/milvus is a vector database that stores and searches vectors (embeddings) in large-scale unstructured data such as text and images. The latest release of this repository is v3.0.3. However, the provided materials do not include a change list for this version, so the version label is all that can be confirmed.
The implementation languages are Go and C++, and it supports horizontal scaling and real-time streaming updates, built on CPU and GPU hardware acceleration, a fully-distributed architecture, and a K8s-native architecture. Standalone mode for single-machine deployments and Milvus Lite, which offers a quick start in Python, are provided as well, and the license is Apache-2.0. The repository was created on September 16, 2019, and currently has 46,453 stars.
That said, public benchmarks and figures on performance are not present in these materials. Milvus Lite is described only as an entry point for a quick start in Python, and the materials do not answer how far large-scale service configurations have been validated. Of the scores provided, only the momentum index is filled in at 4.7, while the technology, quality, and novelty scores are empty. The star count shows user scale only and does not guarantee quality, so it can be viewed as a reference for learning about deployment options, but the current materials are insufficient grounds for deciding on production adoption.