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

2 comments

• fesens 40 minutes ago
  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).
• fabiofachini92 36 minutes ago
  Amazing!