Recent advances in fitting prompt emission spectra in gamma-ray bursts (GRBs) are boosting our understanding of the still elusive origin of this radiation. These progresses have been possible thanks to a more detailed analysis of the low-energy part ($<$,100,keV) of the prompt spectrum, where the spectral shape is sometimes found to deviate from a simple power-law shape. This deviation is well described by a spectral break or, alternatively by the addition of a thermal component. Spectral data extending down to less than 1,keV are extremely relevant for these studies, but presently they are available only for a small subsample of {it Swift} GRBs observed by XRT (the X-ray telescope, 0.3-10,keV) during the prompt emission. The space mission th will allow a systematic study of prompt spectra from 0.3,keV to several MeV. We show that observations performed by th will allow us to discriminate between different models presently considered for GRB prompt studies, solving the long-standing open issue about the nature of the prompt radiation, with relevant consequences on the location of the emitting region, magnetic field strength and presence of thermal components.