Compare an agent version against its baseline on a test suite, and see exactly which cases got worse.
ProWeb and MobileMotion
Preview
Part of Pro, with every other Pro component and every icon. See pricing.
Related
Models Dual Run ViewProComponentRun one prompt on two models side by side and vote blind before the names are shown.Knowledge BaseProComponentSources that ground an assistant, each with a visible ingestion pipeline and a way to test what it retrieves.OrchestratorProComponentA lead model splits a goal into a plan, runs helpers in parallel, then merges what they found.Agent TasksProComponentBackground coding tasks: a composer with repo, branch and model, filters, and rows with status, diff, checks and one action.