Skip to content


Welcome to the Kùzu docs!

Kùzu is an embedded graph database built for query speed and scalability. It is optimized for handling complex join-heavy analytical workloads on very large graphs, with the following core feature set:

  • Supports property graphs and automatic mapping of RDF datasets to property graphs
  • Cypher query language
  • Embedded (in-process) integration with applications
  • Columnar disk-based storage
  • Columnar, compressed sparse row-based (CSR) adjacency list/join indices
  • Vectorized and factorized query processor
  • Novel and very fast join algorithms
  • Multi-core query parallelism
  • Serializable ACID transactions

If you’re new to Kùzu and graph databases, we recommend starting with this introductory video. Once you’re ready to dive in, check out the rest of the documentation!

Why Kùzu

Although there are many graph database management systems (GDBMSs) in the market today, Kùzu stands apart because it addresses specific trade-offs via its design and implementation that make it a compelling choice for analytical query workloads on large graphs. Below, we list some of the key reasons why you may want to consider using Kùzu.

Performance and scalability

Kùzu is a state-of-the-art graph DBMS and is built by a core team of database experts who spent many years doing academic research. Its design and implementation are based on a relentless focus towards scalability and performance, with the founding engineers having co-authored several cutting-edge technical papers and articles on managing and querying large-scale graphs.


Kuzu is built for industry, and implements a suite of features that lower the barrier of entry to modeling your records as a graph, so you can query those records in an expressive graph query language (Cypher). Because Kùzu is an embedded database, it runs in-process with your application, allowing for simplicity in setup with no external dependencies and no hassles in managing DBMS servers.


Kùzu is designed to be highly interoperable with a variety of external formats and columnar or relational stores, including Parquet, Arrow, DuckDB and more. This allows you to easily move your existing data to and from Kùzu, making it a great choice for graph data science, machine learning and analytics use cases.

Structured property graph model

The data model in Kùzu is based on the property graph model, with the addition of structure (including node/edge tables and a pre-defined schema). This makes it flexible and intuitive to model your existing data as a graph, while also being strict enough to optimize query performance and perform vectorized operations at scale.

Open source

Kùzu is open source and has a permissive MIT license, allowing you to easily get started building your commercial and proprietary applications on top of it. While you check out the repo and try out Kùzu, consider giving us a star and spreading the word!