© 2015, Springer-Verlag Berlin Heidelberg. The characterization of the accuracy of ionospheric models currently used in global navigation satellite systems (GNSSs) is a long-standing issue. The characterization remains a challenging problem owing to the lack of sufficiently accurate slant ionospheric determinations to be used as a reference. The present study proposes a methodology based on the comparison of the predictions of any ionospheric model with actual unambiguous carrier-phase measurements from a global distribution of permanent receivers. The differences are separated as hardware delays (a receiver constant plus a satellite constant) per day. The present study was conducted for the entire year of 2014, i.e. during the last solar cycle maximum. The ionospheric models assessed are the operational models broadcast by the global positioning system (GPS) and Galileo constellations, the satellite-based augmentation system (SBAS) (i.e. European Geostationary Navigation Overlay System (EGNOS) and wide area augmentation system (WAAS)), a number of post-process global ionospheric maps (GIMs) from different International GNSS Service (IGS) analysis centres (ACs) and, finally, a more sophisticated GIM computed by the research group of Astronomy and GEomatics (gAGE). Ionospheric models based on GNSS data and represented on a grid (IGS GIMs or SBAS) correct about 85 % of the total slant ionospheric delay, whereas the models broadcasted in the navigation messages of GPS and Galileo only account for about 70 %. Our gAGE GIM is shown to correct 95 % of the delay. The proposed methodology appears to be a useful tool to improve current ionospheric models.