Each rubric criterion is labeled Climb (harder criteria where frontier models still fail often) or Regression (baseline competence). For regression there is little separation between models at the top, with the leading seven models all between 95% and 97%, while Climb separates them, ranging from 28% to 57%.
Each rubric criterion is labeled Climb (harder criteria where frontier models still fail often) or Regression (baseline competence). For…
Each rubric criterion is labeled Climb (harder criteria where frontier models still fail often) or Regression (baseline competence). For regression there is little separation between models at the top, with the leading seven models all between 95% and 97%, while Climb separates them, ranging from 28% to 57%.
Artificial Analysis · X
Content notes
Original publication and rights belong to the source.