Time to give GLM 5.3 a proper try

Always fun adding another model to the stack

Public benchmarks look solid, so I'm running the same tasks through GLM 5.3, Opus 5 and GPT 5.6 Sol to see how they compare

Will update with my thoughts