NAM @ h78 looks better, but remember its still the NAM at range. Although I will say most of the improvements were made at initialization, which makes me think that the drops are probably positive data.
Also fwiw, some subjective model reanalysis from 48 hours: https://forecastwatch.com/2026/01/27/major-u-s-snowstorm-forecast-accuracy-analysis-january-23-26/
Good to know everyone blew it for us
https://efisher828.github.io/weathernext/
Here's the only public source of weathernext 2 maps I believe. Like 12 hours behind and only ptype, rate, and 500mb heights.