Fable 5.1 costs 6x more than GPT-5.6 Sol in Three.js test

rohanpaul_ai · x · 2026-09-02

In a Three.js scene generation test on AIMLAPI, Anthropic's Fable 5.1 cost $5.69 compared to GPT-5.6 Sol's $0.88. Fable spent its budget on scene density (e.g., chandelier crystals, water waves) for realism, while Sol maintained a consistent low-poly style. However, both made basic logic errors: Fable rendered a pool without water, and Sol simplified scenes to the point of losing key geometry. This highlights the trade-offs between object count, physical plausibility, consistency, and inference cost in visual system generation.

Related event: Fable 5.1 costs 6x more than GPT-5.6 Sol in Three.js test(2 posts)→

Original post →

More from Models

Models channel →