Capability · Framework — rag

Weaviate

Weaviate is a feature-rich open-source vector database. It supports hybrid search (BM25 + vector with fusion), multi-modal and named vectors, tenant isolation at scale, generative search modules that plug in OpenAI/Anthropic/Cohere/Google models, and backup/replication. Weaviate Cloud provides a managed service, and first-party apps like Verba showcase the stack. The 1.x line has been production-grade for years.

Framework facts

Category: rag
Language: Go (Python / TS / Go clients)
License: BSD-3-Clause
Repository: https://github.com/weaviate/weaviate

Install

pip install weaviate-client
# run server:
# docker run -p 8080:8080 -p 50051:50051 cr.weaviate.io/semitechnologies/weaviate:latest

Quickstart

import weaviate
import weaviate.classes as wvc

client = weaviate.connect_to_local()
collection = client.collections.create(
    name='Article',
    vectorizer_config=wvc.config.Configure.Vectorizer.text2vec_openai()
)
collection.data.insert({'title': 'Hello world'})
res = collection.query.near_text(query='greeting', limit=2)

Alternatives

Qdrant — Rust-based open-source alternative
Pinecone — managed alternative
Milvus — cloud-scale open-source
pgvector — Postgres extension

Frequently asked questions

What does 'hybrid search' mean in Weaviate?

Hybrid search combines BM25 keyword scoring with vector similarity using a fusion algorithm (reciprocal rank fusion by default). This usually beats pure vector search on RAG tasks where exact-term matches matter — names, error codes, numbers.

Can Weaviate generate answers directly?

Yes — via its generative search modules. You query with near_text and ask for a generated summary, and Weaviate calls your configured LLM (OpenAI, Anthropic, Cohere, Google, local) with the retrieved objects. Useful for quick RAG demos.

Sources

Weaviate — docs — accessed 2026-04-20
Weaviate — GitHub — accessed 2026-04-20