Any good AI can help with the first draft, but ranking depends more on the quality of the final content than the tool itself. Human editing and original value still matter the most
3.5 is still meta? why not 5? do you have any specific prompts/skills that you use? ever considered using an orchestrator through claude code to have sonnet subagents generate the sections in parallel? any recs in general?Everyone says it is all about the editing which is true to an extent, but some models are definitely easier to work with. Claude 3.5 sonnet is way ahead of gpt-4o for actual readability and flow. If you just use standard prompts though it still looks like ai. You need to build a solid outline first and generate it section by section instead of asking for a 3000 word article in one go... that always ends up being 90% fluff.