I'm also trying to learn to do this, but on a higher scale, with python also for automation. Instead of doing one single prompt that do like 1K words, doing like 10–12 prompts of 100–200 prompts that consolidate a full 1K article within the same topic. I would say that there are multiple options depending on your purpose.
1. One is using the same GPT model, to first generate an outline, and then execute shorter prompts with each element of the outline to compose the whole article. Consider that OpenAI doesn't charge you for the prompts, but for the total tokens generated.
2. Another idea, is querying a PPA API with the original topic of the prompt, to keep generating ideas going around the original topic.
3. The third option I believe that it could be achieved by using the already mentioned, the 32K model, but I don't have access to it, still I have access to the gpt-3.5-turbo-16k model. Still, I don't like gpt-3.5 so I'm currently sticking to gpt-4 with one of the previous two options I've said.
View attachment 281759