AFAIK that model is pretty old, and it was explicitly trained for SVG generation. For other models the capability of generating SVGs of real stuff is accidental. Same as GPT-5.x and Sonnet 4.5+ being able to generate MIDI music.