Building an AI-based semantic search or Q&A application? You might need is a vector database to store and search your embeddings easily.
As part of my experiments with creating embeddings for AI semantic search, I have been collecting and trying out various Vector databases.
This guide lists the best vector databases I have come across so far.
I am not recommending any particular one currently – the list is in alphabetical order and the order doesn’t represent my preference in any way.
Table of Contents
Chroma
Chroma is a new AI native open-source embedding database.
DeepsetAI
Deepest is not a vector database itself but a complete semantic search pipeline in one solution. You can plug in models and other vector databases in it.
Has open source as well as a managed cloud version
Faiss by Facebook
Not a vector database but a library for efficient similarity search and clustering of dense vectors. It’s open source.
Milvus
Milvus has an open-source version that you can self-host.
Also has a free trial for the fully managed version.
pgvector
pgvector is an open-source library that can turn your Postgres DB into a vector database.
Pinecone
Pinecone Has a free limited plan. After that it’s usage-based.
Supabase
Supabase is a managed Postgresql solution that implements storing embeddings using the pgvector extension.
Qdrant
Qdrant is an open-source vector database that is free to use in self-hosted mode. They also have a fully managed cloud version too.
Vespa
Vespa is a product from Yahoo. It’s available both as Open Source Download and as a managed Cloud solution.
Weaviate
Weaviate Has a free open-source version that you can self-host
Or you can use their cloud-based version. Free plan available. After that it’s usage-based
Conclusion
If I missed anything, let me know and I will include it after trying it out.