Zvec: Lightning-Fast In-Process Vector DB from Alibaba
Discover Zvec, Alibaba's open-source vector database that embeds directly into your apps with zero server setup. Search billions of vectors in milliseconds, support dense/sparse embeddings, hybrid search, and run anywhere—from notebooks to edge devices. Latest v0.3.0 adds Windows/Android support, RabitQ quantization, and C-API for AI agents. Install via pip or npm and start building RAG apps today with this production-grade, lightweight powerhouse boasting 9.3k GitHub stars.
Zvec: Alibaba's Lightning-Fast In-Process Vector Database
In the world of AI and machine learning, vector databases have become essential for handling embeddings from LLMs, image recognition, and recommendation systems. But most solutions require complex server setups. Enter Zvec—Alibaba's game-changing, open-source vector database that runs directly in your application process.
🚀 Why Zvec Stands Out
With 9.3k GitHub stars and built on Alibaba's battle-tested Proxima search engine, Zvec delivers:
- Blazing Speed: Search billions of vectors in milliseconds
- Zero Servers: Embed directly—no config, no deployment hassles
- Cross-Platform: Linux, macOS, Windows (new!), Android
- Dense + Sparse: Native support for both vector types
- Hybrid Search: Combine similarity with structured filters
Latest v0.3.0 Highlights (April 2026)
The recent release brings enterprise-grade features:
- Windows (MSVC) & Android support with official Python/Node.js packages
- RabitQ quantization for memory efficiency
- CPU Auto-Dispatch for optimal SIMD performance
- C-API for custom language bindings and AI agent integration
⚡ Get Started in 60 Seconds
import zvec
# Define schema
schema = zvec.CollectionSchema(
name="example",
vectors=zvec.VectorSchema("embedding", zvec.DataType.VECTOR_FP32, 4),
)
# Create & insert
collection = zvec.create_and_open(path="./zvec_example", schema=schema)
collection.insert([
zvec.Doc(id="doc_1", vectors={"embedding": [0.1, 0.2, 0.3, 0.4]}),
zvec.Doc(id="doc_2", vectors={"embedding": [0.2, 0.3, 0.4, 0.1]}),
])
# Lightning-fast search
results = collection.query(
zvec.VectorQuery("embedding", vector=[0.4, 0.3, 0.3, 0.1]),
topk=10
)
print(results)
📦 Simple Installation
Python (3.10-3.14): pip install zvec
Node.js: npm install @zvec/zvec
🎯 Perfect For
- RAG applications (Retrieval-Augmented Generation)
- AI agent memory systems
- Recommendation engines
- Image/video similarity search
- Edge device deployments
🚀 Performance That Scales
Zvec handles 10M+ vectors with sub-10ms query times. Check their comprehensive benchmarks for detailed results across hardware configs.
🌐 Join the Community
- Discord for real-time support
- GitHub: alibaba/zvec
- Docs: zvec.org
Whether you're building production RAG pipelines or experimenting with local embeddings, Zvec eliminates vector search complexity while delivering enterprise performance. Fork it, star it, and start embedding today!
Ready to try? pip install zvec and search millions of vectors in your next Jupyter notebook.