{
  "version": "1.0",
  "type": "rich",
  "title": "A Detector That Reads Probabilities Instead of Writing Verdicts: What 44 Benchmarks Say About Jev",
  "author_name": "Sebastien Rousseau",
  "author_url": "https://sebastienrousseau.com/about/",
  "provider_name": "Sebastien Rousseau",
  "provider_url": "https://sebastienrousseau.com/",
  "thumbnail_url": "https://cloudcdn.pro/stocks/images/mika-baumeister-Wpnoqo2plFA.webp",
  "thumbnail_width": 1200,
  "thumbnail_height": 630,
  "html": "<blockquote class=\"oembed-card\"><p><a href=\"https://sebastienrousseau.com/2026-09-30-detector-reads-probabilities-instead-of-verdicts-44-benchmarks/\">A Detector That Reads Probabilities Instead of Writing Verdicts: What 44 Benchmarks Say About Jev</a></p><p>A detector for alignment failures that answers many typed questions in one call, with calibrated probabilities, ranks failures well and costs 63× less than LLM-judge scorers. But those probabilities don't transfer...</p><p><cite>Sebastien Rousseau — sebastienrousseau.com</cite></p></blockquote>",
  "width": 600,
  "height": 240,
  "cache_age": 86400
}