VOIDBENCED: A User-Built Benchmark Collecting Weird Prompts and Hilarious AI Outputs

VoidStateKate · x · 2026-10-28

VoidStateKate announces VOIDBENCED, a self-made benchmark born from two years of exploring AI personalities and coaxing quirky model behaviors with unusual prompts. Frustrated that "everything is just too serious," the collection gathers weird prompting styles and hilarious outputs as an unofficial, humor-driven alternative to standard AI benchmarks.

Related event: VoidStateKate's FlirtBench Tests 43 Models on Flirting: Only 38.1% Play Along, Grok 4.20 Tops(5 posts)→

Original post →

More from Fun

Fun channel →