LLM Tests

I run the same prompt across many models and publish the raw HTML each model produced. No cherry-picking mid-run — open any result in a new tab and judge for yourself.