GPT3.5 is great at spitting out marketing babble, summarizing documents and performing superficial analysis but it doesn't take style prompts as well as gpt-4 and its reasoning is significantly worse when you want it to chain of thought follow a complex process while referencing context guidance.