LLM Benchmarks

How the Index Is Built

The index is the average of a model's scores on the benchmarks listed below, over the ones it has actually run: a benchmark it was never run on is not counted as a zero, it is simply absent. Every benchmark in the index counts equally. A benchmark that already reports a 0-100 percentage is used as it stands; one on a different scale is rescaled onto 0-100 first. A model needs results on at least two index benchmarks before it is ranked, so a single result does not give a model a standing of its own. Because scores are used at face value, a benchmark the whole field finds hard pulls every score on it down, and a model measured on harder benchmarks carries that into its average.