- IXP release notes
- Communications Mining release notes
- Unstructured and complex documents release notes

IXP release notes
Manage Taxonomy page enhancement
We have enhanced the Manage Taxonomy page so you can make inline edits and save updates in bulk as a single new version instead of multiple ones.
Version history for field groups and fields
The new version history feature for field groups and fields tracks their configuration and instructions historically across versions. In addition, it allows you to compare older versions with current ones, and restore to previous states as needed. The feature is accessible through the Manage Taxonomy page and when you edit fields and field groups in the annotation pane. For more details, check Managing the taxonomy and Modifying instructions.
Model-based default attribution option
Model-based is now the default attribution option for new IXP Unstructured documents projects.
Gemini default extraction model
Gemini is now the default extraction model for new IXP Unstructured documents projects.
Bug fixes
Some documents would be stuck in the Saving draft status after you submitted annotations.
Bug fixes
When the system sorted documents in Build, it displayed the document numbering incorrectly in the annotation view.
New API audit events
New API audit events are available and cover project creation and deletion, model updates for publishing and unpublishing model versions, as well as updating model tags. For more details, check API audit events.
Improved header
We have simplified the header to create a cleaner, more intuitive user experience. All IXP-specific functionalities are now grouped together within the IXP sub-header. These functionalities include the project selector and access to Administration when you are in a project.
Bug fixes
The document view would sometimes scroll up or shift focus when you annotated fields in the annotation pane, or when you validated predictions or edited values in the table view.
Bug fixes
- When you imported a taxonomy into a new project, an Exact Text field type would duplicate unnecessarily.
- Model-based attribution caused some references for correctly predicted values to be incorrectly returned as Missing.
- Documents in the annotation view did not respect the sorted or filtered order of documents displayed in the document table in the Build page.
- The warning for outdated predictions in the annotation view incorrectly persisted after new predictions were returned.