Ask HN: How do the OpenAI Insert/Edit functionalities work?

Over a year ago OpenAI released insert and edit endpoints for GPT (<https://news.ycombinator.com/item?id=30689902>). But I can't find any information anywhere about how these features actually work.

My best guess is that it's a fine-tuning of the base Instruct-GPT model combined with some search algorithm, but is there any information about what that search algorithm is? Has anyone reproduced the functionality with any other models?

For reference, this is the functionality I'm referring to: Insert <https://platform.openai.com/docs/guides/completion/inserting-text> Edit <https://platform.openai.com/docs/guides/completion/editing-text>