On the Warwick Model that was shown yesterday, Pagel pointed out they thought they were following a particular part. Like any model though, now we're later on in the process we can look at the assumptions, and if the assumptions turned out invalid, then the conclusions have to also be.
www.medrxiv.org/content/10.1101/2021.12.30.21268307v1.full.pdf
One of the biggest points on the forecasting outcome is
Throughout we have assumed that Omicron has the same generation time distribution as Delta - essentially the same latent and infection periods. However, the rapid increase of Omicron relative to Delta could partially be due to a shorter generation time; Omicron would still need to have a competitive advantage over Delta but this would be magnified by a shorter generation time. As such, if the generation time of Omicron was half that of Delta (so around 2.5-3 days instead of approximately 5-6 days), once the model is recalibrated to match the growth of SGTF, this would approximately halve the predicted peak outbreak sizes
We now have a lot of confidence that the generation time for omicron is less - although I've not heard a prediction that specifically says it's managed to be halved, but either way, their model would say lower peak than the graph offered.
They also assume length of hospital stay is the same between delta and omicron, however hospitalisation -> in patient ratios so far have come down considerably implying a shorter average stay here, and explicit evidence on shorter average stays is available.
Another thing of note, their model predicts no change in wave size if NPI's are introduced on the 4th of Jan, so no reduction from locking down.
As you know, I don't have much truck with modellers (not because they don't do a good job, it's just too hard a job where small assumptions cause huge impact!) but that's good news, all the assumptions I could find would change their results towards a smaller wave.