AI-slop blind test data and reproducible scripts land on GitHub

PawelHuryn · x · 2026-09-07

Pawel Huryn published the AI-slop blind test's raw data, scripts and results to his GitHub experiments repo. Everything is reproducible: exact scripts, unedited logs, and READMEs with method, sample size and caveats; the repo also hosts prior experiment sets like bug-hunt-bench, five-models-three-harnesses, and stealth system-prompt fingerprinting. He asks if followers want it as a live benchmark like Bug Hunt Bench.

Related event: Blind Test of 12 Models: Gemini 3.8 Most 'AI-Flavored'(3 posts)→

Original post →

More from Models

Models channel →