Gemini 3.1 PRO...pure disappointment

A hands-on comparison of Google's Gemini 3.1 Pro vs Claude Opus 4.6 across nine real-world coding and UI generation tasks. Despite strong benchmark scores, Gemini 3.1 Pro consistently underperformed: it only won on flat visual design in the first two tests, then fell behind on 3D rendering, interactive simulations, stock tracking apps, and a circuit board toy. Key issues included tool-calling failures, incomplete feature implementation (marking sections 'coming soon'), poor environment detection (npm vs bun), and inability to one-shot complex prompts. Claude Opus completed most tasks in a single prompt with superior functionality, animations, and attention to detail.