Kùzu v0.13.6 Released: Critical Bug Fixes and Performance Enhancements for the In-Memory Graph Database
Once these steps are complete, your existing component validation will continue to work as before, now powered by Pydantic.
Utilizes a highly optimized columnar storage format tailored for rapid scans and graph traversals.
For Python users, the updated wheel can be pulled directly from PyPI: pip install kuzu==0.13.6 Use code with caution. Basic Usage Example
or complex data pipelines, these "under-the-hood" fixes mean fewer interruptions and more reliable multi-core parallelism when joining billions of edges. Get the Latest: You can upgrade your environment immediately: pip install kuzu --upgrade Use code with caution. Copied to clipboard kuzu v0 136 fixed
Capable of handling graphs with hundreds of millions of nodes and billions of edges.
Kùzu v0.1.3.6 introduces more aggressive memory deallocation and better buffer manager coordination during the copy process. This ensures that the system stays within its allocated memory limits even when processing millions of nodes and rels. 2. Cypher Query Parser Refinement
: Fixed small memory leaks tied to intensive, concurrent read/write operations. Extension Reliability
kuzu v0 136 fixed (primary), Kuzu v0.136 benchmark, upgrade Kuzu, Kuzu memory leak fix, Kuzu concurrency patch, Kuzu JSON parser, Kuzu migration guide. Kùzu v0
If you are experiencing a problem with Kùzu, especially one potentially related to a version, here is a practical troubleshooting guide:
an existing graph project to Kùzu, or are you starting a fresh implementation with this new version?
MATCH (a:Person)-[:KNOWS*1..5]->(b:Person) WHERE a.id = 136 RETURN b.name
Kuzu stores data in a graph structure, comprising nodes (or vertices) and edges, which is particularly useful for modeling complex relationships between data entities. Basic Usage Example or complex data pipelines, these
Implementation of fsync() on Write-Ahead Logs (WAL) during database opening and fullfsync support specifically for macOS/iOS to prevent data corruption.
However, the open-source community has stepped in. The MIT-licensed source code is being actively carried forward under a new name: . For FreeBSD users, for example, the databases/kuzu port is being updated to Ladybug. If you are starting a new project, it is highly recommended that you search for "Ladybug database" to ensure you are building on the actively maintained version of the software.
If you are building graph-based applications—from recommendation engines to fraud detection—staying current with these "fixed" releases is essential for maintaining data integrity and query performance. What is Kùzu?
What kind of (e.g., GNN training, RAG knowledge graphs) are you running? Share public link