Kuzu V0 136 Fixed =link= Link

In October 2025, Kùzu Inc. announced that it was "working on something new" and officially the Kùzu repository. While the core team is no longer actively supporting the project, the code remains open-source under the MIT License . What this means for you:

| Metric | Kuzu v0.135 (unstable) | Kuzu v0.136 (fixed) | Improvement | | --- | --- | --- | --- | | Average query latency | 340 ms | 212 ms | | | Memory usage (peak) | 5.2 GB | 1.8 GB | 65% reduction | | Multi-threaded throughput | 1,200 ops/sec | 2,450 ops/sec | 104% increase | | Crash rate (24 hours) | 1 crash per 6 hours | 0 crashes | 100% stable | kuzu v0 136 fixed

For those new to the ecosystem, Kùzu is designed for query speed and ease of use. It implements the query language and is built to handle large-scale graph datasets directly within your application process (similar to SQLite but for graphs). Its primary strengths lie in its columnar storage architecture and vectorized query execution engine. The v0.1.3.6 Update: What’s Been Fixed? In October 2025, Kùzu Inc

kuzu --version # If output shows "0.136" without a build hash, you are affected. What this means for you: | Metric | Kuzu v0

: Significant work went into casting between unions and fixing issues with nested unions and union arrays, providing more flexibility for semi-structured data.

MATCH (a:Person)-[:ACTED_IN]->(m:Movie)<-[:ACTED_IN]-(b:Person) WHERE a.name = 'Alice' AND a.ID <> b.ID RETURN DISTINCT b.name;