It is. But I noticed something weird. If your prompt is “A cartoon of XYZ” and press enter the preview will be a cartoon but the other images will be weird realistic ones.

The preview is using a different faster model so you're not going to get the exact same styles of responses from the larger slower one. If you have ideas on how to make the user experience better based on those constraints please let us know!

Well my feedback would be that your larger slower model doesn’t seem to be capable of generating cartoon style images while the preview model does seem to be able to.