Thermal permafrost models based on conductive heat transfer are usually forced by time series of surface or air temperatures, as well as snow depth. Examples are GIPL2 developed at the University of Alaska Fairbanks, CryoGrid 2 by University of Oslo. For a more complete representation of land-surface coupling, we have included the surface energy balance as upper boundary of a thermal permafrost model based on CryoGrid 2. This new permafrost model, denoted CryoGrid 3, is for instance capable of reproducing strong near-surface temperature inversions, as they are frequently found under polar night conditions. Furthermore, infiltration of melt- and rainwater in the snowpack is included in CryoGrid 3. We present preliminary results of CryoGrid 3 on modeled ground temperatures, as well as on the radiation budget and turbulent fluxes for validation sites on Svalbard and Northern Siberia. CryoGrid 3 is designed as a flexible platform, which can be adapted and extended with relative ease to include further permafrost processes, such as ground subsidence and thermokarst formation through melting of excess ground ice, or heat transfer processes in Arctic lakes and ponds. Thus, CryoGrid 3 can serve as a development tool for additional process modules in the land-surface schemes of atmospheric models.