I find that GPT-4 always wins in task specific, clear prompts. E.g. I used it on a project recently to translate code from one version to another, and keep in

I find that GPT-4 always wins in task specific, clear prompts. E.g. I used it on a project recently to translate code from one version to another, and keep in mind that the syntax was very different A comparable example would be translating old class based React code to functional TSX with hooks. And it does extremely well there. Other coding tools can’t even do that as there’s no way to load a whole prompt and the entire original file as a prompt.