Same prompt significantly different results

API Feedback
,
1

I have used the same prompt for several days on gpt4o receiving what I needed 9 times out of 10 and used up all daily usages multiple times.

Fed up of waiting, I paid for some credits and reused the same prompt on the API. So far I’ve generated 50+ images and not one is what I need. They are bad.

It’s like they’re completely different models (GPT’s Dalle 3 and API’s Dalle 3)

Why is there only 1 Dalle 3 model listed? They are obviously not the same. Or parameters are not available to tune them to be the same. I won’t buy more credits because the API version is bad as-is.

2

For information I’ve also confirmed the prompts are identical by clicking on the image in GPT4 and the (I) and reusing the same prompt.

I’ve also tried style and quality.