Kuzu V0 136 Full ((hot)) Jun 2026

that combines semantic vector search with traditional keyword search. Implementation Guide: Hybrid Search Feature Define Schema with Vector & Property Support

Kuzu's columnar storage and vectorized query processor enable it to handle multi-join queries—common in graph analytics—far faster than row-based systems. 2. Perfect for RAG

: Uses columnar disk-based storage and vectorized/factorized query processing to handle complex, join-heavy workloads.

The days of wrestling with complex SQL joins or heavy server-based graph databases are over. With Kuzu, you can bring the power of graph analytics directly to your application code, making complex data relationships simple to query and manage. kuzu v0 136 full

CALL CREATE_HNSW_INDEX('Document', 'embedding', 'dist=COSINE'); Use code with caution. Copied to clipboard Full-Text Search Index

| Item | Details | |------|---------| | | Kuzu – a high‑performance, embeddable graph database written in C++ with a Rust‑friendly API and a Python front‑end. | | Version | 0.13.6 (the first “full” release after the 0.13 series of incremental builds). | | Release date | 12 March 2026 | | License | Apache 2.0 (permissive, commercial‑friendly). | | Target audience | Data engineers, data scientists, and developers who need fast graph analytics on‑the‑fly, especially in AI‑augmented pipelines. | | Core promise | Sub‑microsecond query latency on billion‑edge graphs while keeping the footprint under 300 MB in RAM. |

: This version continues to refine its HNSW vector index and full-text search , making it a powerhouse for RAG (Retrieval-Augmented Generation) and AI applications. Perfect for RAG : Uses columnar disk-based storage

The COPY FROM command now handles:

Improved integration for cloud-native deployment scenarios.

Kuzu v0.1.36 continues to operate as a single library with no external dependencies. It can be embedded directly into C++, Python, Node.js, or Java applications. This removes the need for Docker containers or separate server processes, drastically lowering the barrier to entry for application developers. faster startup times

Node Index: [ 0 ] [ 1 ] [ 2 ] │ │ │ CSR Offset: ▼ ▼ ▼ Array: [Edge1, Edge2] [Edge3] [Edge4, Edge5, Edge6] (Contiguous memory blocks ensure high CPU cache hits) Factorized and Vectorized Execution

Unlike older graph systems that store nodes and edges as scattered, semi-structured JSON-like blobs, Kùzu organizes data into strict tables.

Simplifies DevOps, faster startup times, runs in the same process. Expressive and standard query language for graph data. Performance