Opus 4.6 currently leads the remote labor index at 4.17. GPT-5.4 isn't measured on that one though: https://www.remotelabor.ai/

GPT 5.4 Pro leads Frontier Maths Tier 4 at 35%: https://epoch.ai/benchmarks/frontiermath-tier-4/

> We're gonna need some new benchmarks...

You can't consistently benchmark something that is qualitative by nature. I'm struggling to understand how people don't understand this.

Humanity's Last Exam (HLE) is already insanely difficult. It introduces 2,500 questions spanning mathematics, humanities, natural sciences, ancient languages, ...

Here is an example question: https://i.redd.it/5jl000p9csee1.jpeg

No human could even score 5% on HLE.