BenoStreamDB is distributed as an embedded library for Python and Rust. It runs completely locally with no separate server required.
Install the official Python package via PyPI. Precompiled wheels for Linux, macOS, and Windows (x86_64 and ARM64) with DataFusion SQL, PyArrow, CUDA, and WGPU support.
Add the core library to your Rust project via crates.io. Cross-platform embedded engine with direct memory access and zero-copy Arrow arrays.
Integrate BenoStreamDB with your existing data stack using these open-source connectors. Every link below resolves to v0.16.0, the same release this documentation was built from.
Precompiled DataSource V2 JARs for Spark 3.5 & 4.0 (Java 17 & 21) with embedded native engine.
The alias above is the Spark 4.0 / Java 17 build. Every combination is published for this release as spark-benostream-spark-{3.5,4.0}-java-{17,21}.jar.
Precompiled plugin package for Trino 482 & 484. Query BenoStreamDB tables directly through Trino.
The alias above is the Trino 484 build. Per-band packages are published for this release as trino-benostream-{482,484}.zip.
Transform data in BenoStreamDB using dbt models, tests, and documentation with the dbt-benostreamdb adapter.
High-performance columnar data transfer over gRPC. JDBC/ODBC/ADBC drivers connect any SQL client.
Connect AI agents and LLMs to BenoStreamDB via the Model Context Protocol. Run instantly via uvx or pip.
OpenSearch-compatible REST API and Qdrant vector search endpoint. Drop-in replacement for Elasticsearch 7.10 and Qdrant clients.
BenoStreamDB is Serverless by Default, but if you need to access it over the network, you can run the built-in Arrow Flight SQL server or the OpenSearch-compatible REST Gateway. Check the Documentation for setup instructions.