gpt-6IMAGE

Muse, Opus 5.5 and Astra image outputs compared on the same prompt

The author shares image results from three models on the same task and says GPT-6 Astra looks like it has "zero taste," though a few days of real testing are needed to tell whether it is only benchmark-tuned.

Angel 🌼

@Angaisb_Archived 15 days agoOriginally posted 2026.10.13

Muse, Opus 5.5 and Astra image outputs compared on the same prompt (image 1 of 3)

Muse, Opus 5.5 and Astra image outputs compared on the same prompt (image 2 of 3)

Muse, Opus 5.5 and Astra image outputs compared on the same prompt (image 3 of 3)

Angel 🌼 @Angaisb_Originally posted 2026.10.13

Open original on X

Still, we'll have to try it and see how it works for people over the next few days to know if it's as good as it looks or it's just benchmaxxed From what I've seen so far, it seems to have zero taste Left: Muse Wonder (max effort), middle: Opus 5.5, right: GPT-6 Astra

This post quotes another post: x.com/angaisb_/status/2109689406663643532

x.com/angaisb_/status/2109700462660420029Archived 2026.10.13 10:12