HWE Bench: A new unbounded Benchmark for LLMs (GPT 5.5 is on top)

Current benchmarks have ceilings, usually 100%. This benchmark aims to be a long lasting, high correlation with the ability to solve real world problems and follow complex instructions, and unbounded (meaning it can always go higher).

Very nice!!