I expect every company to try to make a model as good as they possibly can, especially now that Mythos has served as a proof of concept to demonstrate that there's lots of interest in AI for cybersecurity. But if they don't try, that hardly assuages concerns about not being able to access the very best models, does it?