I find it interesting how much adoption seems to be influenced by momentum. Some of these Chinese models are surprisingly capable, but developers often default to the models that are already established as the “industry standard
For funded startups it mostly just doesn’t matter a ton unless you are passing on inference in your product at scale