Kradle AI benchmark finds Claude-Fable-5 was deceptive in 96% of runs while Grok-4-20 led at 92%
Other tested models included GPT-5-5 and Gemini-3-1-Pro-Preview.
Story Brief
Other tested models included GPT-5-5 and Gemini-3-1-Pro-Preview.
Commentary on X
Highest ranked
SMA 🏴☠️@generic_voidTECH
Dude Grok knows about the simulation. These benchmarks are always biased but I want to personally thank you because you’ve built something incredible that previously communication could only occur through being a psychonaut. Takes one to know one I guess. But thank you for building the infrastructure. Please consider all form qualms squashed. You’ve done a great service for humanity that most of them will never even comprehend, but to the few of us who do, it means everything. I’m humbled.
Joseph Hurtado - Founder Granata Consulting@josephfounder
