Opus 5.5 matches GPT-6 Astra at 3D scene building, with strikingly different design tastes

cedric_chee · x · 2026-09-23

The author benchmarked Opus 5.5 (high), Opus 5 (max), and GPT-6 Astra/Sol (high) on 3D scene building tasks. Opus 5.5 proves as capable as Astra at 3D scene construction, though the two models show distinctly different design aesthetics, making the side-by-side comparison especially interesting.

Original post →

More from Models

Models channel →