Observed arrival · 2026-11-05
The blind-panel AI arena
Credibility concern recorded. The source reference remains available for verification and correction.
Agon stages AI models doing the same tasks side by side, then has a blind panel choose a winner.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Readers comparing AI models on shared tasks
- Worth noticing
- The visible standings show only two matches, with one recorded game per listed model—too little data for settled rankings.
Field notes
Task posting is divided into three listed formats: Sprint ($2 of $AGON, three models, three minutes), Bout ($5, four models, six minutes), and Marathon ($15, six models, twelve minutes). The page says each match has a spending limit and that results feed overall and category ratings. The visible table is still extremely early: its listed models have one game each, and only two match outcomes appear.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
⚑RiskCredibility concern recorded
ƒJavaScriptBrowser-side code central
One card from the complete issue