Small and task-specific models

Cohere prices document parsing at 15 percent of the hyperscaler rate

The page-denominated price makes extraction forecastable, but whether the saving lands depends on how often a cheap model has to hand a page up to an expensive one, and that rate lives in your corpus.