Hi Ruud,
You are comparing a hydrostatic model (GFS) with a non hydrostatic model (ICON). A hydrostatic model does not include the level of physic modules that the non-hydrostatic models have and this affects the area of resolving convection. Differences in boundary layer physics will also affect the resolving of gusts. This is not true in regards to WRF which can resolve gusts being a non-hydrostatic model (see the attached images).
In the attached examples you can see the GFS and also Arpege (Global) are very soft in gusts behind the front line. These are hydrostatic models
The images for openWRF, ICON, ICON_EU and Arpege_EU all show significantly more gust activity behind the front. These are non-hydrostatic models
It is not always just about resolution. There are different types of models that have different limitations.
By the way, GFS will be going non-hydrostatic when the new FV3 core becomes operational in the near future.
David