Dalle-3 in ChatGPT wrong output and various issues

Currently when I try running multiple sets, and ok I get it its bussy and a lot of people are using it, but it now also started churning out ‘semi random’ prompts that are similar, after the first set, but completely different then the premise I was working with.

I guess this will not last forever, but other then that I also notice even though I set the canvas size and landscape mode in the beginning at certain intervals it reverts to the basic square at random points throughout the chat and as it doesn’t take those prompts but alters them (this used to work when I first started using it) so you can’t replicate them in the format you want, I find it weird too that if you can actually get it you can create the image in a different format and it will be the exact same image.

I guess it is set to be variable in output, but for some reason Dall-E in all facets of its use is inconsistent, from veering really hard away from the premise, to inconsistent output amounts, often no explanation about causes of bugs, etc.

It seems also currently it lost all ability to track what I did before and to make realisations or derive new context and meaning based on the previous, images are stopping and consistently although good looking nonesense images not related to anything I asked for.