Every AI API, priced from the source.

14 providers, input and output price per million tokens, each figure read off the vendor's own page and stamped with the date it was read — catalogue pass 2026-08-04, with high-change examples rechecked separately. Treat each page's date as the evidence boundary.

Pricing is provider evidence. Runtime performance is a different evidence class: the benchmark page publishes the protocol and withholds latency or quality rankings until a reproducible run is complete.