What method did you use to generate that one?
I tried using their OpenAI-compatible API and got one that's a lot less fun (no animation) - https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
What method did you use to generate that one?
I tried using their OpenAI-compatible API and got one that's a lot less fun (no animation) - https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
I used the web UI at https://chat.qwen.ai/. I didn’t log in. I selected Qwen3.8-Max-Preview and prompted “create an SVG of a pelican riding a bicycle.”
It took a LOT of reasoning time before it spit it out. 10 min or more?
I wish I had created an account first because I lost the chain-of-thought that it used. I remember a couple highlights. There was an internal monologue about whether it was worth it to animate the legs and pedals, and a lot of repeated text about focusing on “a cohesive visual design.”
I don’t make svg art so I’m no expert, but I am also surprised at how small the resulting svg ended up. I expected LLM output to be more verbose, especially considering the animations.
I’m curious what happens if you retry a couple times. The nondeterminism might be strong enough to have highly variable outcomes.