Was just testing out our assistant with the “upgraded” model and it takes significantly longer and the results were much worse… wondering if it needs a complete different style of prompting or something. What have you guys seen with the new model?
It does need a bit of adjustment in how you prompt. Not sure if you caught the email we sent, but please check out our updated prompting guidance for o3-mini: https://platform.openai.com/docs/guides/reasoning#advice-on-prompting. Can you make some adjustments based on this and let us know how you fare?
1 Like
Awesome thank you. Yes just actually read through the full email and saw that. Have made adjustments and am seeing it get much closer to what we have with gpt4o but still not as good. Will keep tweaking over the weekend