Is anyone aware of any device characterisation work being done to improve to he device models for SKY130? Are those the models they provide to customers?

There is a set of so-called "continuous" models (which is a permanent PR on the Google repo and is merged into the Efabless repo) but they are just a re-hashing of the existing characterization data, and it's not clear that they are an improvement other than that I think they are continuous across W and L for monte carlo simulation, which the initial set of devices definitely is not. The "continuous" models are the ones from the s130 PDK version 2.1.302 given to customers, but I don't know if those are what's in the current s130 version. They are missing a lot of devices and I had to go to a lot of trouble to put together a "combined" set of models and parameters that started with the "continuous" models and folded in the rest of the missing devices from sky130.