Currently FME can read 12dxml file format (.12dxml), but cannot directly read the zipped version of this file (.12dxmlz).
This is similar to the difference between a Google Earth KML file and a KMZ file.
A current work around is the routine can have a user parameter for file name, it starts with a creator, uses the "ZipExtractorAnyExtension" from FME hub to extract the file, then the file path can be used in a feature reader with the format set to 12dxml.
There are two issues with the workaround - it requires a user parameter to be setup within the routine, and it requires unzipping the file contents to a folder prior to being usable in the routine. 12dxml files can be large compared to 12dxmlz (7mb vs 206kb) so file size is significant, especially when 12dxmlz files get into 100mb's+.
That is why a dedicated 12dxmlz reader would be great.