nodebb-plugin-openai doesn't use streaming responses it sends the post to openai and once the response is received it makes a reply in the topic or chat.

@josef said in [nodebb-plugin-openai] NodeBB OpenAI Plugin: Now you just need to add a Retrieval Augmented Generation (RAG) so that GPT can answer based on the existing content in the forum! It's actually a bit problematic, because NodeBB keeps an indexes for search but not a vector version (embedded)

One obvious ChatGPT use case is summarizing an entire thread. It's very simple with most of the LLM APIs. It would just need to be run when the thread has changed significantly – there's some probably some nuance on that front.