db = kuzu.Database("my_graph_db") conn = kuzu.Connection(db)
Bindings for these languages have been updated to support the new vector data types, ensuring you can pass float arrays from your application layer directly into the query engine.
: Data transfer happens directly in memory. kuzu v0 120
: Queries are processed in batches using CPU SIMD instructions to improve cache locality and multi-core parallelism.
Kùzu distinguishes itself through an "embedded" architecture similar to DuckDB, but optimized for graph relationships: Vectorized Processing db = kuzu
Released on October 10, 2025, v0.120 introduces several critical updates that expand the database's analytical and integration capabilities. A. Graph Algorithm Extensions
The v0.1.20 release's stability and performance make it an ideal foundation for such agentic systems, as the database engine is capable of handling complex, join-heavy analytical queries that are common when traversing knowledge graphs. : Improved performance for in-memory HNSW graphs by
: Improved performance for in-memory HNSW graphs by compressing neighbor offsets, significantly reducing the memory footprint for high-dimensional vector search.