Same prompt.
Different results.

7 AI models did the same everyday tasks five times each, once with a vague prompt and once with a precise one. Every output is here, graded, with the reason it passed or failed.