Model Showdowns · New matchup every week
Same prompt. Different AI. Touch the results.
Every week we feed one identical prompt to competing AI models and publish exactly what each one builds — raw, unedited, fully interactive. No benchmark charts, no cherry-picking. See the differences yourself, then steal the prompt and rerun the test.
● Live nowInteractive app
The Planetarium Test
One prompt asked two models to build a working planetarium from first principles — real orbital math, no libraries, no data files. Both shipped a working night sky.
Build a complete planetarium as a single HTML file. Compute the live positions of the Sun, Moon, and five visible planets from Keplerian orb…
Claude Opus 4.8
Max effort
VS
Claude Fable 5
Max effort