OpenAI has an Edit endpoint but it has a very limited rate limit and its unclear if that will ever be raised. Also my attempts to communicate with OpenAI suppor

OpenAI has an Edit endpoint but it has a very limited rate limit and its unclear if that will ever be raised. Also my attempts to communicate with OpenAI support about related matters were generally ignored for a month before receiving any response. So that's why it outputs the full code. It may be possible to build a diff-based solution also but its a bit complicated and when I tested with normal diff the previous models were not able to do it consistently. Allowing users to generate from the homepage would require anonymous generation which I am hesitant to allow now because it would be unpaid usage and no username to pass to the API, and OpenAI may ban accounts that are associated with certain prompt content. But its definitely a good suggestion to consider to consider anyway.