As a fun and timely benchmarking exercise, I gave the following prompt to GPT 5.6 Sol (High), Opus 5 (High), Gemini 3.1 Pro (Extended Thinking), Gemini 3.6 Flash, Kimi K3 (via OpenRouter) and Muse Spark 1.1 (Thinking). All of them were one-shotted barring Gemini 3.6 Flash which gave me a broken image as the artifact and I had to ask it to fix it.

“I want a lovely visual way to immerse myself with The Odyssey by Homer. I just saw the movie by Chris Nolan and loved it. Map all the key events in the epic poem to locations in real world, and let me take Odysseus' journey on my own in a fun and interactive visually appealing way. I want to post this on my Squarespace site so it needs to be something that works as a single html file but not lose all the awesomeness and delight. Ground it in both the poem and what we know.”

For the thumbnail images, I used the following prompt.

“can you give me a single image that reflects this html page?”

Which one did you like the best?