Different ai models drawing a cat with only python turtle

A viral social media experiment has put four leading AI models—Qwen3.5-9B-Q8, DeepSeek Thinking, Claude Sonnet 4.6 Extended, and Gemini 3.1 Pro—to a creative coding test. The user gave each model the identical prompt: "Write a python turtle program that draws a cat." The goal was to compare how different architectures and training approaches handle a simple, open-ended programming task with a visual output. The results, shared on Reddit, show four distinct and complete Python programs, each producing a unique graphical cat using the Turtle graphics library.

Each model's solution revealed its "personality." Some generated highly structured, modular code with separate functions for drawing ears, eyes, and whiskers, while others produced a more linear, script-like approach. The visual cats varied significantly in style, complexity, and artistic interpretation, from simple line drawings to more detailed silhouettes. This test moves beyond standard benchmarks, offering a tangible, side-by-side comparison of how these models approach problem-solving, code generation, and even basic creative instruction. It highlights differences in how they interpret ambiguity, structure logic, and implement Python libraries.