GFS is one model out of five or so mainstream NWP solutions.
The 00Z MOGREPS suite for instance keep it cold until the 18th.
I'm not saying GFS will be wrong, but all the focus over one model in it's extended range isn't healthy.
You are, of course, spot on, but happens all the time because of both the plethora of data available for each run and the fact that it runs every six hours.
Realism is required: how often do we get an uninterrupted cold spell lasting 3 weeks, which would be required to get us through the Xmas weekend? Some sort of breakdown is likely, but what it looks like, whether it’s temporary or not, and what happens and where at any breakdown are all completely up in the air.
The focus needs to be on the upcoming 4-5 days, with that comparative rarity of cold surface and cold upper temperatures and slack low pressure. As for the here and now, the thermometer shows it’s still below freezing here (after -3c overnight) the radar returns show lots of activity in the predicted places.
Location: South Cambridgeshire
130 metres ASL
52.0N 0.1E