Because of the time-lagged ensemble approach used by the Met Office vs. the "burst" approach used by ECMWF, the January C3S update of GloSea5 only had a max. ~40% probability of a major SSW, versus 100% from SEAS5. Very different tropospheric forecasts likely a result of this!
Charts available at https://climate.copernicus.eu/charts/c3s_seasonal/
Here's a comparison of the different C3S ensemble systems from https://confluence.ecmwf.int/display/CKB/Description+of+the+C3S+seasonal+multi-system. DWD, CMCC and ECMWF launch all their ensemble members on the 1st of the month - so can deterministically predict shorter-term processes, like the 5th Jan sudden warming.
Taking UKMO as an example - the forecast consists of ensemble members launched throughout the preceding month (in this case, December) as a different way of generating spread. The major SSW was not deterministically predictable at those lead-times, so the signal is weaker.
This is why making the distinction between the "model" and the "prediction system" is important. Here, the type of system used (lagged ensemble) is probably the cause behind the different (potentially worse? - we shall see!) tropospheric forecast, rather than the actual model.