We critically examine recent claims of a high solar metallicity by von Steiger & Zurbuchen (2016, vSZ16) based on in situ measurements of the solar wind, rather than the standard spectroscopically inferred abundances (Asplund et al. 2009, hereafter AGSS09). We test the claim by Vagnozzi et al. (2016) that a composition based on the solar wind enables one to construct a standard solar model in agreement with helioseismological observations and thus solve the decades-old solar modelling problem. We show that, although some helioseismological observables are improved compared to models computed with spectroscopic abundances, most are in fact worse. The high abundance of refractory elements leads to an overproduction of neutrinos, with a predicted 8B flux that is nearly twice its observed value, and 7Be and CNO fluxes that are experimentally ruled out at high confidence. A combined likelihood analysis shows that models using the vSZ16 abundances are worse than AGSS09 despite a higher metallicity. We also present astrophysical and spectroscopic arguments showing the vSZ16 composition to be an implausible representation of the solar interior, identifying the first ionization potential effect in the outer solar atmosphere and wind as the likely culprit.