Compare models.
Put benchmarked models side by side: composite score, cost, tokens, and how they do across each domain. Only models from the same track and task set are compared, because a composite is only directly comparable within one cohort.
Pick models to compare
Choose two or more models on the leaderboard - use “Add to compare” on each card - and they line up here side by side. Models can be compared only within the same track and task set.
Back to the leaderboard