Recently, I’ve been doing a lot of copywriting comparison tests between GPT-5.2 and Gemini 3 Pro. Conclusion: The gap in natural flow has narrowed, and GPT-5.2 no longer feels so 'science-y.' When GPT-5.2 first launched, I still leaned towards using Gemini 3 Pro for tasks like text polishing and structural rewrites because 'natural flow' is something that’s hard to judge as clearly as factual accuracy or logical correctness. But after some time (admittedly intensive but informal testing), my personal experience is that ChatGPT-5.2 Thinking has indeed improved in readability, pacing, and the ability to distill key points. Especially when you include some clear writing goals in the prompt, it can more consistently produce versions that 'feel human-written and are easy to read.' At least, it’s starting to meet some of my requirements. Here’s a practical little tip: just add specific writing goals directly into the prompt, like: > More engaging to read, higher information density, clearer logic, fewer technical terms, feels like a viral tweet but stays objective. This alone can significantly improve the quality of the output. After a few rounds of fine-tuning and optimization, you can achieve the final draft requirements.