Community Rankings
AI Model Rankings
Rankings are derived only from eligible votes tied to completed comparisons. Empty or unavailable data stays visibly empty or unavailable.
How the rating is calculated
Forge Atlas expands each eligible vote into pairwise ELO outcomes. A chosen winner is scored against every other completed model in that comparison. A tie records pairwise ties. Ratings start at 1200 with K=32. Preference ratings measure community preference; they are not scientific benchmark scores.