Modal

No quality score appears here until it is real. When this provider is scored, it runs the same task bank as every ranked row: Static Facts, Dynamic Facts, Multi-Source, and Deep Research. Judges are pinned and calibrated against human labels first, with the agreement statistics published. See the methodology.