Category
Vector databases: what they are and who makes them
Pinecone, Qdrant, Weaviate and 5 more, sorted into types, with the company behind each and where it is based.
Types of vector databases
Managed-only services
Pinecone is the clearest case: no self-hosted edition exists, and you pay for storage, reads and writes on a hosted index. You trade control for little operations work. Weaviate Cloud and Qdrant Cloud sit nearby but also ship open-source code you can run elsewhere.
Open-source engines you can host
Qdrant, Weaviate, Milvus, Chroma and LanceDB publish their code on GitHub. You can start on a laptop or a container, then move to the vendor's cloud or stay on your own servers. That keeps an exit open but puts patching and scaling on your team.
Distributed and lakehouse-style systems
Milvus is built as a distributed, Kubernetes-native system for very large collections, and LanceDB stores vectors beside raw multimodal data in an open table format. Both suit teams whose data volume, or whose model training pipelines, have outgrown a single-node database and need storage built for scale.
Vector search inside a general database
MongoDB Atlas adds vector and hybrid search to the same cluster that holds operational data, which removes a sync job. This route suits teams already on that database, though a dedicated engine may give finer control over index settings and cost at scale.
Facts at a glance
| Tool | Publisher | Based in | Access |
|---|---|---|---|
| Pinecone | Pinecone Systems, Inc. | United States | Free tier + API |
| Qdrant | Qdrant Solutions GmbH | Germany | Free tier + API |
| Weaviate | Weaviate B.V. | Netherlands | Free tier + API |
| Milvus | LF AI & Data Foundation project | United States | Free tier + API |
| Zilliz Cloud | Zilliz Inc. | United States | Free tier + API |
| Chroma | Chroma | United States | Free tier + API |
| LanceDB | LanceDB | United States | Free tier + API |
| MongoDB Atlas Vector Search | MongoDB, Inc. | United States | Free tier + API |
