Arena 实测：Gemini 4 Pro 水机物理胜过 Opus 5.5
同一提示词做 3D 水上飞机浮力与水面交互模拟，Arena 预览版 Gemini 4 Pro 据称压过已发布的 Claude Opus 5.5。
Gemini 4 Pro (in the arena) outperformed Claude Opus 5.5 in a 3D seaplane physics simulation. That says more about Google's trajectory than any single benchmark could. Same prompt, both models: build a 3D floatplane simulation with working physics. Gemini 4 Pro reportedly nailed the buoyancy and water interaction. Opus 5.5 did not keep up, according to the arena comparison making the rounds. The kicker is that Gemini 4 Pro has not even officially shipped yet. This was a preview build beating a production model on a task built specifically to stress-test spatial reasoning. My honest take: one demo does not crown a champion, and floatplane physics is a narrow slice of what these models actually get used for. But if Google is closing gaps like this before launch day, Anthropic does not have the comfortable lead people assumed a few months ago. The model wars just got a lot less predictable. Does this change who you trust for serious coding work, or is this still just a highlight reel?
这条是引用着另一条写的：「Arena 对比：Gemini 4 Pro 对 Claude Opus 5.5 做 3D 水上飞机物理模拟」