The GOLF instrument on board SoHO has been in operation for almost 25 years, but the ageing of the instrument has now strongly affected its performance, especially in the low-frequency pressure-mode (p-mode) region. At the end of the SoHO mission, the ground-based network BiSON will remain the only facility able to perform Sun-integrated helioseismic observations. Therefore, we want to assess the helioseismic performances of an échelle spectrograph such as SONG. The high precision of such an instrument and the quality of the data acquired for asteroseismic purposes call for an evaluation of the instrument's ability to perform global radial-velocity measurements of the solar disk. Data acquired during the Solar-SONG 2018 observation campaign at the Teide Observatory are used to study mid- and low-frequency p modes. A Solar-SONG time series of 30 days in duration is reduced with a combination of the traditional IDL iSONG pipeline and a new Python pipeline described in this paper. A mode fitting method built around a Bayesian approach is then performed on the Solar-SONG and contemporaneous GOLF, BiSON, and HMI data. For this contemporaneous time series, Solar-SONG is able to characterise p modes at a lower frequency than BiSON or GOLF (1750 μHz versus 1946 and 2157 μHz, respectively), while for HMI it is possible to characterise a mode at 1686 μHz. The decrease in GOLF sensitivity is then evaluated through the evolution of its low-frequency p-mode characterisation abilities over the years: a set of 30-day-long GOLF time series, considered at the same period of the year from 1996 to 2017, is analysed. We show that it is more difficult to accurately characterise p modes in the range 1680 to 2160 μHz when considering the most recent time series. By comparing the global power level of different frequency regions, we also observe that the Solar-SONG noise level in the 1000 to 1500 μHz region is lower than for any GOLF subseries considered in this work. While the global p-mode power-level ratio is larger for GOLF during the first years of the mission, this ratio decreases over the years and is bested by Solar-SONG for every time series after 2000. All these observations strongly suggest that efforts should be made towards deploying more Solar-SONG nodes in order to acquire longer time series with better duty cycles.