Comment by simonw

It's a test of text-based LLMs to see how good they are at SVG geometry. Video models are a different category of software entirely.