Semantic Models Lab Release notes¶
Semantic Models Lab provides the foundation for the Semantic models feature.
Version 1.1.0 - May 29th, 2026¶
Compatible with DSS 14.4.0 and above.
Semantic Model Editor¶
Fixed an issue preventing to use “Extract from Document” feature in the glossary
Fixed deletion of golden queries
Removed auto save to prevent save errors and confusion:
Users must now manually save their work
Running the webapp in a container is now working properly
Version 1.0.0 - May 12th, 2026¶
Compatible with DSS 14.4.0 and above.
New feature: Cross-entity metrics¶
This feature is only compatible with DSS 14.6.0 and above.
The Semantic Model Editor now lets you define metrics that combine attributes across multiple entities or joins. The Semantic Model query tool can then use these metrics during SQL generation when the user question matches their intent.
Cross-entity metrics can be used to model business calculations such as revenue, margin, or conversion logic when they depend on data spread across several entities.
New feature: Snowflake import¶
Snowflake Semantic Views can be imported using the Semantic Model Editor. A Dataiku DSS Snowflake connection is required. Snowflake tables referenced by the imported Semantic Views must already exist in your DSS instance so they can be used during the import.
Querying¶
Semantic model and Dataset SQL query tool settings have been enhanced (project and dataset / semantic model selection)
Fixed truncated records for final answer generation
Fixed dataset reference handling during the export of projects with semantic models
Updated SQL Generation parameters:
Max query rows: Determines the number of rows the LLM will have access to in the SQL query output
Max rows in sources: Determines the amount of rows in the sources
Max rows in artifacts: Determines the number of rows generated in the artifacts
Query LIMIT adjusted to be the higher number of the above three parameters max(max_query_rows, sources_rows_limit, artifacts_rows_limit)
Semantic Model Editor¶
Added a search input to the entity list
Fixed UI and state change issues
Playground¶
Added labels explaining SQL generation parameters
Version 0.2.0 - Apr 3rd, 2026¶
Compatible with DSS 14.4.0 and above.
New feature: Single dataset query tool¶
This new tool allows querying a single dataset, without having to build a complete semantic model first. The tool automatically builds a temporary semantic model on the fly.
New feature: Simple execution¶
Both the Semantic Model Editor, Semantic Model query tool and Single dataset query tool can now either use either a simplified or full (agentic) orchestration methodology. Previous versions only supported the full orchestration.
Full orchestration provides better results for complex semantic models, but requires more back and forth with the LLM, which means that execution is overall slower. The simple orchestration provides good results in most cases, and is significantly faster.
Querying¶
Both query tools now support propagating end-user identity to query the underlying databases
Generated SQL limits are now properly preserved when enforcing row caps
Fixed multiple edge cases in configuration
Semantic Model Editor¶
The entity creation workflow has been reworked
Fixed issue with entity id normalization causing save failures
Multiple UI issues have been fixed in the editor
Version 0.1.0 - Feb 9th, 2026¶
Initial release of Semantic Models Lab for DSS 14.4.0