Indeed, a reduced prompt size is unlikely to compensate for 8x inference cost increase. Also fine tuning OpenAI's black box with data generated using gpt4 does

Indeed, a reduced prompt size is unlikely to compensate for 8x inference cost increase. Also fine tuning OpenAI's black box with data generated using gpt4 does not really require a lot more skills than prompt engineering.