GLM-4 Thinking

Best for deep reasoning tasks - complex architectural refactors.