The building stock accounts for 43% of the total French final energy consumption in 2017. The French building stock mainly includes residential and tertiary buildings, which have very different consumption patterns. Today, building stock models are largely used in many countries for energy demand forecast and decision making. In this context, good knowledge of the building stock is essential for modeling. However, this type of model is lacking for the tertiary sector, as it is difficult to categorize buildings that have a wide variety of area and energy uses, different systems and complex dynamic trends. Existing stock models are mostly based on an aggregated representation of the total heated floor area of the building stock. Such a representation limits explicit studies of the factors affecting the energy demand and fails at taking into account heterogeneity. This paper exposes and discusses new methods aiming at modeling the tertiary building stock at the city scale in France. While mainly relying on open source public data, methods are designed to be applicable at any scale (from small cities to large region or even national scale).