chroma

Install chromadb, create a client and collection, then call collection.add with documents, metadatas, and ids. Use collection.query with query_texts and n_results to run similarity search, optionally adding a where clause for metadata filters.

Chroma is open-source and runs locally, suiting development and self-hosted projects. Pinecone is a managed cloud service with auto-scaling, while FAISS provides pure similarity search without metadata storage or filtering.

Yes, use chromadb.PersistentClient with a path argument to persist data to disk. Data is saved automatically and can be reloaded later by connecting with the same path.

Yes, Chroma integrates with both frameworks. Use langchain_chroma.Chroma as a vector store or retriever in LangChain, or ChromaVectorStore with a StorageContext in LlamaIndex for index-based querying.

Pass a where clause to collection.query or collection.get with operators like $gt, $gte, $lt, $lte, $ne, $in, and logical combinators $and and $or. For example, where={"category": "tutorial"} restricts results to matching documents.

Avoid Chroma when you need managed auto-scaling cloud infrastructure, where Pinecone fits better, or when you need high-performance production workloads, where Qdrant or Weaviate are more suitable.