Recent studies have suggested a mismatch of up to 0.20 dex between iron abundances derived from Fe I and FeH lines in the H-band spectra of M dwarfs and in this work we investigate the nature of this possible offset. We analyze near-infrared H-band APOGEE spectra of stars in 18 binaries composed of a G dwarf primary and an M dwarf secondary, together with four M dwarf stars having measured angular diameters from the literature, and six M dwarf members of the Hyades and Coma Berenices open clusters. These three families of benchmarks were used to evaluate the FeH line list and check for possible systematic uncertainties in the FeH gf-values. Our tests used 1D local thermodynamic equilibrium plane-parallel model atmospheres, a radiative transfer code, and the baseline APOGEE spectral line list to derive metallicities for the binary G dwarf primaries using Fe I lines, while stellar parameters and metallicities for the M dwarfs used both FeH and Fe I lines. The mean metallicity obtained for the Hyades M dwarfs was <[Fe/H]> = +0.08 ± 0.04, and for Coma Berenices it was <[Fe/H]> = +0.02 ± 0.08. The metallicities of the G and M dwarfs in binary systems showed excellent agreement (0.06 dex), and the mean metallicities for the open clusters were also consistent with literature values from high-resolution optical analyses. We investigated the consistency between the spectroscopic and interferometric Teff scales, finding agreement within the uncertainties. Forcing full agreement between the spectroscopic and interferometric Teff scales resulted in a poorer match for the metallicities in the binaries and the open clusters. We conclude that the best overall concordance is obtained when no adjustments are made to the FeH gf-values that are based on the R. J. Hargreaves et al. line list.