libSQL

Overview

libSQL is an open source, open contribution fork of SQLite, created and maintained by Turso. It's production-ready, fully backwards compatible with SQLite, and adds features like native vector search capabilities.

Vector Search Features

• DiskANN Algorithm: Implements Approximate Nearest Neighbors (ANN) using the DiskANN algorithm for efficient vector indexing
• LM-DiskANN Variant: Uses the LM-DiskANN algorithm optimized for scenarios where the dataset is too large to fit in memory
• Native Integration: Vector indexing is built directly into the database, eliminating the need for separate vector databases

How to Use

Creating a vector index:

CREATE INDEX movies_idx ON movies (libsql_vector_idx(embedding));

Searching with the vector_top_k virtual table:

SELECT * FROM vector_top_k('movies_idx', query_vector, k);

Performance Characteristics

While slower than in-memory approaches due to disk access latency, the FreshDiskANN implementation minimizes these delays as much as possible. It's particularly well-suited for:

• Edge computing environments
• Mobile applications
• Multi-tenant architectures where memory usage must be kept low

Use Cases

• Semantic search in SQLite applications
• Embedding storage and retrieval on edge devices
• AI applications requiring vector similarity search
• Scenarios requiring both relational and vector data in one database

Pricing

Free and open-source.