I gave Claude, Codex, and Antigravity nothing but screenshots of an app, and one rebuilt it almost pixel for pixel
www.makeuseof.com/gave-claude-codex-and-antigravity-nothing-but-screenshots-of-an-app/
I deliberately chose an app the models were unlikely to already know. Familiarity would have ruined the test
Codex was the best all-rounder. It made one hilarious mistake, though
Claude did a decent job. It got creative where I really didn’t want it to
Antigravity had no major flaws, but it still came last. It looked good, just not close enough