Observed arrival · 2026-09-08
Keibu, the Language Model With No Privacy
Credibility concern recorded. The source reference remains available for verification and correction.
Keibu puts questions to a language model in a public queue, then permanently publishes both the prompts and its answers.
Field notes
Keibu processes submissions through a shared queue backed by a single computer, whose sleep and wake states appear in a public event log. The page reports 12 answers, including a 411-token response that took 70.3 seconds, and displays refusals alongside ordinary replies. Free submissions are limited to six per hour while the queue is clear; a 0.001 SOL payment can move a question ahead, while $KEIBU holdings accrue another form of access credit.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue