As Ray notes, it depends...
Following is likely TMI!
For my 200 point model, it took 88'. This is based on the time delta between first and last image, which is all captured in the model file (*.PNT) in C:/Program Data/AstroPhyscis/APPCC/Models (if you do not change the default folder). The actual time is a bit longer to initialize SGP and APPM and load the model after.
About 25" to take a 5" image (19 M full-frame), plate solve, slew to next position, 2" settle time. This can vary, mostly with the slew time. I use SGP to take the image and plate-solve either with ASTAP or a local install of Astrometry.net. ASTAP is much faster, but does not always solve. AN is slower, but has never failed, if I set it up correctly. Fairly fast PC.
If you go back through threads from last 6 months or so, you will find a discussion of doing localized sub-models in the areas you plan to image on a given night. A valid approach, but open to discussion as to any time savings. I recall Roland using a technique where he used a filter (H-alpha ?) during the model build that allowed him to start taking images for the model long before it was dark enough to take long-exposure images.