The power of asteroseismology relies on the capability of global oscillations to infer the stellar structure. For evolved stars, we benefit from unique information directly carried out by mixed modes that probe their radiative cores. Large progress has been made for understanding the mixed-mode pattern. I will present the recent findings (period spacings, rotational splittings, coupling factors, mode widths, mode visibilities) and show how they can be used to deliver stringent tests for probing stellar interiors. These results, when expressed in terms of ensemble asteroseismology, help constraining stellar evolution.