What’s New?
Support for Python 3.14. SDV libraries now support Python 3.14. We encourage all users to upgrade to the latest version. Old versions of Python (3.9 and below) are not maintained, which can lead to bugs and security vulnerabilities over time. For more information, see the official status of Python versions.
Please note that all SDV bundles also support 3.14 at this time, except for AI Connectors and SDV Agent. We are working on adding support for these two remaining bundles.
Enhanced quality report. The quality report from our SDMetrics library now specifically checks that the synthetic data meets patterns that are already present in the real data. It no longer gives credit to any weak/non-existent correlations. As a result, you may see a change in the quality score, which paints a more accurate picture of your synthetic data quality.
To learn more about this, visit our Inside the Vault section.
Read CSV data more accurately. Starting from this release, SDV makes sure to keep leading zeros that may appear in your data. For example, if your date is represented in a DDMMYYYY format like “04022026”, it will no longer cut off the leading zero and convert it to an integer. This leads to higher quality synthetic data. For the best experience, we recommending using our CSVHandler to read in your data.
Are you an SDV Community user? Check the community release notes on Github.