Weaviate

Weaviate is an open-source vector database for AI applications. It stores high-dimensional embeddings (from models like OpenAI text-embedding-3 or Cohere Embed) along with metadata and supports vector, keyword, and hybrid search. It's used for retrieval-augmented generation, semantic search, recommendations, and any application that needs nearest-neighbor search over vectors. Weaviate is available as open source (self-hosted) and as a managed cloud service.

Yes. Weaviate is fully open source under the BSD-3 license and can be self-hosted at no cost. You can run it on Docker, Kubernetes, or bare metal with no usage restrictions. Weaviate Cloud Services (WCS) is the paid managed option, starting in the tens of dollars per month for small production indexes and scaling up to enterprise contracts for larger deployments.

Pinecone is managed-only and often the fastest path to production for teams that don't want infrastructure work. Weaviate is open source, which means you can self-host for free or use Weaviate Cloud if you want managed. Weaviate also has stronger hybrid search and a modular architecture that lets you plug in different embedding models and even do in-database generation for RAG. Pinecone has a larger ecosystem and more polished managed experience.

Hybrid search combines dense vector search (based on embedding similarity) with BM25 keyword search (based on exact and near-exact term matching) using a configurable weighting. This matters because dense vectors can miss rare or specific terms (product codes, names, acronyms), while keyword search alone misses semantic similarity. Weaviate's hybrid search is one of the strongest in the vector database market and significantly improves retrieval quality for RAG and enterprise search.

Modules are pluggable components that extend Weaviate's functionality — for example, text2vec-openai for OpenAI embeddings, text2vec-cohere for Cohere, generative-openai for in-database RAG generation, and reranker-cohere for result reranking. You enable the modules you need in your Weaviate configuration, and Weaviate handles calling the external APIs automatically. This modularity is one of Weaviate's biggest advantages over more monolithic vector databases.

Yes. Because Weaviate is open source, you can run it on any infrastructure — self-hosted on-premise, in a private cloud, or air-gapped environments. This is a key advantage for regulated industries (banking, healthcare, defense) that can't send data to a managed vector database. Commercial support contracts are available through Weaviate's enterprise team for self-hosted deployments.

Yes. Weaviate has first-class integrations with LangChain, LlamaIndex, Haystack, and DSPy. It's well supported as a vector store in most Python and TypeScript RAG frameworks, and the Weaviate client SDKs make it easy to use directly if you don't want a framework. Most RAG tutorials that use Pinecone can be adapted to Weaviate with minimal code changes.