Turso is an in-process SQL database written in Rust that maintains compatibility with SQLite across its SQL dialect, file formats, and C API. The project represents a complete rewrite of SQLite in Rust rather than a fork, positioning itself as the next evolution of SQLite with native support for modern database features. The software is currently in beta, with the developers noting that while it powers production applications including Turso Cloud, the Kin AI assistant, and Spice.ai, it has not yet reached SQLite-level reliability standards.
The repository demonstrates active development with a median issue and pull request response latency of 0.0 hours and a mean latency of 142.9 hours across 2433 tracked items. The most prolific contributors are penberg with 2690 events, LeMikaelF with 1999 events, and jussisaurio with 1535 events. Bug reports, correctness issues, and enhancement requests represent the most active issue categories with 408, 259, and 258 labeled items respectively. The project shares overlapping contributors with ClickHouse, DuckDB, and Bun, indicating cross-pollination within the database and systems programming communities.
Core features include BEGIN CONCURRENT for improved write throughput using multi-version concurrency control, change data capture for real-time tracking of database modifications, and asynchronous I/O support on Linux via io_uring. The database supports vector operations including exact search and vector manipulation. Multi-language bindings are available for Go, JavaScript, Java, .NET, Python, Rust, and WebAssembly, enabling integration across diverse technology stacks. The project runs on Linux, macOS, Windows, and browsers through WebAssembly compilation.
Experimental features include encryption at rest for local data protection, incremental computation using DBSP for incremental view maintenance and query subscriptions, full-text search powered by the tantivy library, and multi-process WAL coordination via a .tshm sidecar file for cross-process write-ahead log readers and writers. The roadmap includes vector indexing for fast approximate vector search similar to libSQL vector search capabilities.
The repository includes a Model Context Protocol server mode that allows AI assistants to interact with databases through nine tools covering database operations from opening databases and listing tables to executing queries and modifying schemas. This integration supports Claude Code, Claude Desktop, and Cursor, enabling natural language database interaction.
The project is classified across multiple domains including SQLite compatibility, distributed database capabilities, edge computing, embedded database functionality, serverless operation, data replication, low latency performance, global scale deployment, managed service architecture, and data persistence. The codebase is available under the MIT license and distributed through multiple package managers including crates.io for Rust, npm for JavaScript, PyPI for Python, and Maven Central for Java, making it accessible to developers across different programming ecosystems.
