Gemini 4 vs GPT-6 Astra on the same Spider-Man game prompt
Given the same prompt, Gemini 4 produced a single comic-style image while GPT-6 Astra built a playable 3D city with rooftop traversal and combat.
🚨 GEMINI 4 MADE A COMIC PANEL. GPT-6 ASTRA MADE THE GAME. Same prompt. Same task. Build a Spider-Man Game GEMINI 4 → huge focus on the character → dark, cinematic city backdrop → looks almost like a comic-book page brought to life GPT-6 ASTRA → full 3D city to move through → rooftop traversal → enemies + actual combat → much stronger sense of scale The difference is kind of insane. One interpreted the prompt visually. The other turned it into an actual game. WHICH ONE UNDERSTOOD THE ASSIGNMENT?
This post quotes another post: x.com/spiderman/status/2034225034009415951