SemaDB

Overview

SemaDB provides vector and hybrid search capabilities through two versions: v1 for pure vector search and v2 for multi-index hybrid keyword search with higher overhead but more features.

Key Features

• Quantized Vector Search: Uses quantizers to reduce memory usage while maintaining search quality
• Hybrid Search: Combines vector and keyword search to find relevant documents
• Filter Search: Supports filtering based on queries or metadata
• Simple REST API: JSON or MessagePack based, no custom query language required
• No Custom Clients: Standard HTTP clients work out of the box

Two Versions

Version 1 (v1)

• Pure vector search implementation
• Lower overhead and simpler architecture
• Optimized for semantic similarity search

Version 2 (v2)

• Multi-index implementation
• Hybrid keyword search support
• Higher overhead but more versatile
• Better for combining semantic and lexical matching

API Design

SemaDB features a RESTful API that is:

• JSON or MessagePack based for efficient data transfer
• No need to learn a new query language
• Compatible with standard HTTP clients
• Easy integration with existing applications

Use Cases

• Semantic search applications
• Hybrid retrieval combining keywords and vectors
• RAG implementations requiring metadata filtering
• Applications needing simple REST-based vector search
• Memory-constrained deployments with quantization

Technical Features

• Vector quantization for reduced memory footprint
• Metadata filtering capabilities
• Support for multiple similarity metrics
• Efficient storage and retrieval

Pricing

Open-source with self-hosting options.