Observed arrival · 2026-08-29
The Prompt’s public model fight
A paid public experiment where 25 participants contribute words to one locked prompt, then compare how multiple AI models respond in fresh sessions.
Field notes
The experiment collects one real word or a two-to-five-word phrase from each paid seat, rejects URLs and prompt injections, and closes a round after 25 verified payments. The resulting cumulative instruction is sent without chat history to the listed text models, with the site describing a report for that seat window. Project buyers can attach a website to a race card with tracked visits, while personal participants can remain name-only or add a profile link.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue