LLM canvas coding benchmark — plant-creature vs circles prompt

LLM canvas coding benchmark — plant-creature vs circles prompt

@empiricail shares a single-file HTML canvas prompt (procedurally animated plant intercepting oscillating circles) used to compare ChatGPT/Gemini/Claude on fluid animation + collision logic. Reusable LLM-coding eval prompt.

(no takeaways yet)

ChatGPT, Claude, Gemini API integration, code generation, single-file canvas prompt

Prompt 👇 "Develop a single HTML file featuring a high-speed interactive simulation where a procedurally animated plant creature on the left side of the screen automatically intercepts and consumes circles rushing in from the right with erratic, oscillating vertical trajectories. Use the HTML5 Canvas API to manage complex collision detection and smooth animations, ensuring that the plant’s movements remain fluid and that no circular entities successfully reach the left boundary." This prompt tests the AI’s ability to coordinate a complex interaction between a dynamic, multi-part character and a constant stream of moving targets. It requires the AI to calculate physical intersections in a high-speed environment while simultaneously managing the organic visual movement of the plant. The AI m…