> A model that was released a couple months ago scores 50% higher than SWE-2, a model released today, on an out-of-sample benchmark. Can I say I’ve come out of this more impressed with Sol?
Wait a second, are we taking into account the massive difference in terms of resources of these two companies?