Rotating Box Challenge: Why OpenAI GPT Beat DeepSeek and Qwen2.5 Hands Down

Author(s): Tarun Singh

Originally published on Towards AI.

This member-only story is on us. Upgrade to access all of Medium.

Imagine a rotating box. Inside it, a ball bounces around, striking the walls, defying gravity, and never stepping out of bounds. Sounds simple, right? Well, not for every AI model.

I decided to put three of the most talked-about AI models — OpenAI GPT, DeepSeek, and Qwen2.5 — to the test. The challenge? A simple physics simulation of a rotating box with a bouncing ball. No technical jargon, no pricing debates, no architecture comparisons — just a practical, real-world demonstration of how well these models can execute a given prompt.

Here’s how it went down.

I gave all three models the exact same prompt:

Physics Simulation: Ball in a Rotating Box Create a Python program that simulates a ball moving inside a rotating square box. The simulation should: – Use Pygame (or Pygame+Pymunk) for physics and graphics – Display a square box rotating continuously around its center – Show a ball affected by gravity and collisions – Ensure the ball stays within the box's boundaries – Provide smooth animation and realistic physics behavior Technical Requirements: – Python 3.x with Pygame – 800×800 pixel window – 60 FPS target performance – Basic collision detection… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI