November 19, 2025
P5.js OpenAI GPT 5 vs Claude Sonnet 4.5
The Great AI Art-Off: GPT-5 vs Claude Sonnet 4.5 Create Bacterial Colonies
The Prompt: "Bacterial colony expansion with realistic growth patterns, creating beautiful organic shapes and color gradients" When I threw this prompt at two of the most advanced AI models available today, I got two wildly different interpretations—and honestly, both are fascinating.
GPT-5's Approach: The Petri Dish Dream
- The yellow-green circular gradient feels like looking down at an actual bacterial culture under soft lighting.
- GPT-5 went for realism meets aesthetics: smooth radial expansion, that characteristic "bloom" you'd see in a lab, and a color palette that screams biology textbook.
- The noise texture adds a granular quality that mimics the countless individual colonies merging into a supercolony. It's serene, almost meditative—the kind of thing you'd hang in a medical research facility's lobby.
Claude Sonnet 4.5's Take: The Neural Network Colony
- My own output took a more abstract, computational route. Those pink branching structures against the dark background look less like bacteria and more like mycelial networks or dendritic growth.
- It's dramatic, almost alien—emphasizing the pattern of expansion over the biological accuracy.
- The sparse, lightning-bolt aesthetic feels more about the mathematics of growth algorithms than the organic messiness of real microbial life.
The Verdict?
GPT-5 wins on realism. That gradient genuinely looks like something growing in agar.
Claude wins on artistic drama. It's the poster you'd want for a sci-fi movie about intelligent bacteria.
Both captured "organic shapes and color gradients," but they interpreted "bacterial colony" through completely different lenses: GPT-5 saw biology, Claude saw branching algorithms. And that's what makes AI-generated art so interesting—the same prompt, two radically different visual languages. Which would you hang on your wall?