Sometimes with poorly supported languages the AI gets into a loop. As an example I just created a Bambara story with 16300 words (800 requested) and it was already in a loop around word 115. I guess this is wasteful.
Is it correct that Story Bot always uses the better AI, not mini? That’s what the page says, and in doing some tests it seems that way to me.
Thanks for letting me know! Seems that the model can’t really handle Bambara.
Yes, Story Bot always uses the better AI model (gpt 4o).
(OpenAI does now have a 4.5 model which is apparently better at creative writing, but it’s much more expensive and it’s been deprecated already and is being removed from the API in July)