“Two Four AI Labs is currently working on O-1 style models, reasoning, and test-time computation.”