4 ms·
HWE Bench: A new unbounded Benchmark for LLMs (GPT 5.5 is on top)
- fesens 5mo agoCurrent 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).
- fabiofachini92 5mo agoAmazing!
- paulobeckhauser 5mo agoVery nice!!