DuckDB Labs has released DuckLake 1.0, a production-ready data lake format that stores table metadata in a SQL database to solve the ‘small file problem’ common in object storage. The release introduces features like data inlining for small updates and compatibility with Iceberg-style deletion vectors to improve lakehouse performance. Clients are already available for popular platforms including Spark, Trino, and Pandas, with a roadmap signaling future Git-like branching capabilities.