四模型同提示词建埃菲尔铁塔:Gemini 4 Argon 造型偏粗壮
Kimi K3.1、GPT-6 Astra、Gemini 4 Argon、Opus 5.5 同提示词生成埃菲尔铁塔,Gemini 云朵柔和但塔身偏粗,价格与准确度不成正比。
Four AI models. One prompt: “build the Eiffel Tower.”🗼 The cheapest one made the most convincing tower. Kimi K3.1: $2.04 · 8 mins GPT-6 Astra: $7.45 · 8 mins Gemini 4 Argon: $6.10 · 9 mins Opus 5.5: $8.95 · 10 mins Opus cost over 4x more than Kimi and took longer. What I noticed: → Kimi: the slimmest, most elegant silhouette, with the right taper and proportions → Opus: the richest scene, with the lawns, trees and deep-blue sky → Gemini: soft fluffy clouds, but a stockier tower → GPT: the finest lattice, delicate but a little see-through Price clearly didn’t buy accuracy in this test. Which one looks closest to the real thing?