Imagen 4 Ultra for comparison, 25 seconds for 4 pics at 2k output resolution, which is still quite fast in comparison to 4o spending at least a couple of minutes or little less in optimal conditions. Same simple prompt, I changed only the captions.
>Painterly anime style, a 18 years old Italian woman with wolf ears and tail, short brown hair in bob cut with fringe, emerald green eyes, wearing Roman inspired attire, she's declaiming the introduction of the Aeneid, one hand on her chest for pathos, captioned below in overlaid elegant cursive captions "... Italiam fato profugus Laviniaque venit..." Set on a beach in southern Latium, late afternoon. Peaceful and idyllic atmosphere.

It really likes to give her that white-tipped tail, anyway... I'm also getting the impression that Ultra might try harder to read into "pathos", in comparison to Fast.