MIT Open source & free forever

The vector database built for speed

Blazingly fast similarity search for AI applications. Built in Rust, designed for scale, and completely open source.

⭐ Star on GitHub Read the Docs

Terminal

# Install via cargo

cargo install vectron

# Or use Docker

docker run -p 8080:8080 vectron/vectron:latest

# Start the server

vectron serve --port 8080

⭐12,438 stars

🍴1,847 forks

👥234 contributors

📦v2.4.1 latest

// features

Why developers choose Vectron

Optimized for the most demanding AI and ML workloads.

⚡

Sub-millisecond Search

HNSW indexing with SIMD acceleration. Query billions of vectors in under 1ms.

🦀

Written in Rust

Memory-safe, zero-cost abstractions, and fearless concurrency. No GC pauses, ever.

📐

Multi-Modal

Store and search text, image, and audio embeddings in a single collection.

🔄

Real-time Updates

Insert, update, and delete vectors without rebuilding indices. Always consistent.

🌐

Distributed

Horizontal scaling with automatic sharding and replication. Runs anywhere.

🔌

Client SDKs

Official clients for Python, Node.js, Go, Rust, and Java. OpenAPI spec included.

// quick start

Up and running in minutes

Simple API, powerful results. Here's what it looks like.

Pythonclient.py

from vectron import Client

# Connect to your instance

client = Client("localhost:8080")

# Create a collection

client.create_collection(

  name="articles",

  dimension=1536

)

# Insert vectors

client.insert(

  collection="articles",

  vectors=embeddings,

  metadata=docs

)

JavaScriptsearch.js

import { Vectron } from '@vectron/client';

// Initialize client

const db = new Vectron('localhost:8080');

// Semantic search

const results = await db.search({

  collection: 'articles',

  vector: queryEmbedding,

  limit: 10,

  filter: {

    category: 'tech'

  }

});

console.log(results);

// community

Built by the community

234 contributors and counting. Join us.

Want to contribute? Check out our contributing guide and open issues.

// pricing

Free forever. Cloud if you want it.

Self-host for free, or let us handle the infrastructure.

Open Source

Self-Hosted

$0 forever

Full-featured Vectron, run on your own infrastructure.

All core features
Unlimited vectors
Community support
MIT licensed

Get Started

Frequently asked questions

Is Vectron really free for production use?+

Yes. Vectron is MIT licensed — use it however you want, commercially or otherwise, with no restrictions. The Cloud offering is optional for teams who prefer managed infrastructure.

How does Vectron compare to Pinecone or Weaviate?+

Vectron is built in Rust with a focus on raw performance. In our benchmarks, it's 2-5x faster than alternatives for similarity search. Plus, you can self-host it — no vendor lock-in.

What embedding models work with Vectron?+

Any model that produces vector embeddings — OpenAI, Cohere, Sentence Transformers, CLIP, etc. Vectron is model-agnostic. Just specify the vector dimension when creating a collection.

How do I contribute to the project?+

Start with our contributing guide on GitHub. Good first issues are tagged, and our Discord community is happy to help new contributors. We review PRs within 48 hours.