威尼斯洪水模拟：模型对比

原帖说明

I just finished my personal benchmark testing Gemini 4 Argon and Claude Opus 5.5 on the same harness (Devin Desktop) Opus 5.5 (High) vs Gemini 4 Argon (High) The output is a single HTML file, Three.js only, with zero assets. A floating Venice with acqua alta flooding the piazza, waterfalls, gondolas leaving wakes, pigeons taking off when the bells ring and a full day/night cycle in 60 seconds (sped up to 1.2x), exported to mp4 Opus: ~520K tokens (1 hour) Gemini: ~206.9K tokens (36 minutes) What do you think?

— @Fanfulladev