We report on the results of a comparative analysis carried out on more than ten months of co-located records collected at the J9 gravity station in Strasbourg (France) with a new generation spring gravimeter, the Micro-g LaCoste, Inc. gPhone-054 and the GWR-C026 superconducting gravimeter (SG-C026). The gPhone is essentially a LaCoste & Romberg, model G meter, but with an improved thermal system (a double oven) for increased temperature stability, which should result in an unprecedented improvement in noise and drift of this mechanical gravimeter. We test the performances of the gPhone-054 in terms of resolution, accuracy, noise level and long-term stability (drift) with respect to the SG-C026. Our comparative analysis is performed in a wide spectral domain, ranging from the body tides to the seismic band. This study confirms that the SG has better performances over the whole analysed spectral band. The gPhone-054 instrumental drift observed during this study still remains a critical point preventing the study of long-term gravity changes. In fact, compared with the SG, the drift is large and even non-linear. We observed a drift rate evolution characterized by a decrease from 50 µGal/day to 15 µGal/day, after about 1 month of operation. This makes it hard to distinguish real-time gravity changes from the instrumental drift. We tried to improve the drift modelling by using frequent absolute gravity (AG) measurements, but unfortunately during that time no significant gravity changes have been detected which would have helped us to discriminate short-term drift excursions from real gravity changes. In terms of noise levels, the gPhone-054 turns out to be about 10 times noisier than the SG-C026 at seismic frequencies, while in the tidal band, it is twice as noisy. In between, at periods ranging from 1 h to 6 h, the gPhone-054 is about 3 times noisier than the SG-C026 but performs slightly better than the Scintrex CG5 in terms of noise level and precision.