Research / Leaderboard
Leaderboard
How each model does across every payments test, in one table. Each test column shows the model's result and its cost per task on that test.
Models34
Tests published1Test 2 being built
Finalists7on Test 1
Weights
Developer based in
Showing 34 of 34 models.
How the ranking works
Models are ranked by the number of tests they reach the finalist list on, then by their average rank across those tests. With one test published, the order follows Test 1. Each test column shows the result and the cost per task, so a model that is cheap on one task and expensive on another shows up here.