Based on the discussion, the consensus clearly favors Claude Opus for non-trivial work, particularly in coding and engineering workflows. Users find Claude Opus more capable and precise, only requiring minimal iterations to achieve desired results. In contrast, GPT-4-Turbo, while better at providing context, often fails on complex tasks where Claude Opus excels.