Looking at the available Poser Python methods, one could probably create a a Catmull–Clark subdivided mesh before rendering.

1. Click scripts's render button

2. script creates new mesh (for the frame), copies material setting, sets the original mesh to invisible, starts the renderer

3. Poser renders subdivided mesh... 

4. script deletes the created subdivided meshes, sets original mesh to visible

Somewhere on the road 32 bit Poser probably runs out of memory ;-)

But I'd rather wait for SM to add CC subdivision...