weaviate

README

Weaviate

Weaviate is an open-source, cloud-native vector database that stores both objects and vectors, enabling semantic search at scale. It combines vector similarity search with keyword filtering, retrieval-augmented generation (RAG), and reranking in a single query interface. Common use cases include RAG systems, semantic and image search, recommendation engines, chatbots, and content classification.

Weaviate supports two approaches to store vectors: automatic vectorization at import using integrated models (OpenAI, Cohere, HuggingFace, and others) or direct import of pre-computed vector embeddings. Production deployments benefit from built-in multi-tenancy, replication, RBAC authorization, and many other features.

To get started quickly, have a look at one of these tutorials:

• Quickstart - Weaviate Cloud
• Quickstart - local Docker instance

Installation

Weaviate offers multiple installation and deployment options:

• Docker
• Kubernetes
• Weaviate Cloud

See the installation docs for more deployment options, such as AWS and GCP.

Getting started

You can easily start Weaviate and a local vector embedding model with Docker. Create a docker-compose.yml file:

services:
  weaviate:
    image: cr.weaviate.io/semitechnologies/weaviate:1.32.2
    ports:
      - "8080:8080"
      - "50051:50051"
    environment:
      ENABLE_MODULES: text2vec-model2vec
      MODEL2VEC_INFERENCE_API: http://text2vec-model2vec:8080

  # A lightweight embedding model that will generate vectors from objects during import
  text2vec-model2vec:
    image: cr.weaviate.io/semitechnologies/model2vec-inference:minishlab-potion-base-32M

Start Weaviate and the embedding service with:

docker compose up -d

Install the Python client (or use another client library):

pip install -U weaviate-client

The following Python example shows how easy it is to populate a Weaviate database with data, create vector embeddings and perform semantic search:

import weaviate
from weaviate.classes.config import Configure, DataType, Property

# Connect to Weaviate
client = weaviate.connect_to_local()

# Create a collection
client.collections.create(
    name="Article",
    properties=[Property(name="content", data_type=DataType.TEXT)],
    vector_config=Configure.Vectors.text2vec_model2vec(),  # Use a vectorizer to generate embeddings during import
    # vector_config=Configure.Vectors.self_provided()  # If you want to import your own pre-generated embeddings
)

# Insert objects and generate embeddings
articles = client.collections.get("Article")
articles.data.insert_many(
    [
        {"content": "Vector databases enable semantic search"},
        {"content": "Machine learning models generate embeddings"},
        {"content": "Weaviate supports hybrid search capabilities"},
    ]
)

# Perform semantic search
results = articles.query.near_text(query="Search objects by meaning", limit=1)
print(results.objects[0])

client.close()

This example uses the Model2Vec vectorizer, but you can choose any other embedding model provider or bring your own pre-generated vectors.

Client libraries and APIs

Weaviate provides client libraries for several programming languages:

• Python
• JavaScript/TypeScript
• Java
• Go
• C# (🚧 Coming soon 🚧)

There are also additional community-maintained libraries.

Weaviate exposes REST API, gRPC API, and GraphQL API to communicate with the database server.

Weaviate features

These features enable you to build AI-powered applications:

• ⚡ Fast Search Performance: Perform complex semantic searches over billions of vectors in milliseconds. Weaviate's architecture is built in Go for speed and reliability, ensuring your AI applications are highly responsive even under heavy load. See our ANN benchmarks for more info.

• 🔌 Flexible Vectorization: Seamlessly vectorize data at import time with integrated vectorizers from OpenAI, Cohere, HuggingFace, Google, and more. Or you can import your own vector embeddings.

• 🔍 Advanced Hybrid & Image Search: Combine the power of semantic search with traditional keyword (BM25) search, image search and advanced filtering to get the best results with a single API call.

• 🤖 Integrated RAG & Reranking: Go beyond simple retrieval with built-in generative search (RAG) and reranking capabilities. Power sophisticated Q&A systems, chatbots, and summarizers directly from your database without additional tooling.

• 📈 Production-Ready & Scalable: Weaviate is built for mission-critical applications. Go from rapid prototyping to production at scale with native support for horizontal scaling, multi-tenancy, replication, and fine-grained role-based access control (RBAC).

• 💰 Cost-Efficient Operations: Radically lower resource consumption and operational costs with built-in vector compression. Vector quantization and multi-vector encoding reduce memory usage with minimal impact on search performance.

For a complete list of all functionalities, visit the official Weaviate documentation.

Useful resources

Demo projects & recipes

These demos are working applications that highlight some of Weaviate's capabilities. Their source code is available on GitHub.

• Elysia (GitHub): Elysia is a decision tree based agentic system which intelligently decides what tools to use, what results have been obtained, whether it should continue the process or whether its goal has been completed.
• Verba (GitHub): A community-driven open-source application designed to offer an end-to-end, streamlined, and user-friendly interface for Retrieval-Augmented Generation (RAG) out of the box.
• Healthsearch (GitHub): An open-source project aimed at showcasing the potential of leveraging user-written reviews and queries to retrieve supplement products based on specific health effects.
• Awesome-Moviate (GitHub): A movie search and recommendation engine that allows keyword-based (BM25), semantic, and hybrid searches.

We also maintain extensive repositories of Jupyter Notebooks and TypeScript code snippets that cover how to use Weaviate features and integrations:

• Weaviate Python Recipes
• Weaviate TypeScript Recipes

Blog posts

• What is a Vector Database
• What is Vector Search
• What is Hybrid Search
• How to Choose an Embedding Model
• What is RAG
• RAG Evaluation
• Advanced RAG Techniques
• What is Multimodal RAG
• What is Agentic RAG
• What is Graph RAG
• Overview of Late Interaction Models

Integrations

Weaviate integrates with many external services:

Category	Description	Integrations
Cloud Hyperscalers	Large-scale computing and storage	AWS, Google
Compute Infrastructure	Run and scale containerized applications	Modal, Replicate, Replicated
Data Platforms	Data ingestion and web scraping	Airbyte, Aryn, Boomi, Box, Confluent, Astronomer, Context Data, Databricks, Firecrawl, IBM, Unstructured
LLM and Agent Frameworks	Build agents and generative AI applications	Agno, Composio, CrewAI, DSPy, Dynamiq, Haystack, LangChain, LlamaIndex, N8n, Semantic Kernel
Operations	Tools for monitoring and analyzing generative AI workflows	AIMon, Arize, Cleanlab, Comet, DeepEval, Langtrace, LangWatch, Nomic, Patronus AI, Ragas, TruLens, Weights & Biases

Contributing

We welcome and appreciate contributions! Please see our Contributor guide for the development setup, code style guidelines, testing requirements and the pull request process.

Join our Slack community or Community forum to discuss ideas and get help.

License

BSD 3-Clause License. See LICENSE for details.