How AI pricing actually works
AI APIs charge per token — the little chunks models read text in, roughly ¾ of an English word. You pay for tokens in (your prompt and documents) and tokens out (the answer). This calculator estimates input cost, which dominates most document-heavy workloads.
The spread is the story: the cheapest frontier-class model costs around 1–3% of the most premium one. That’s why serious teams route bulk work to cheap models and save the premium ones for the calls where quality pays. Want the full picture? Start with our plain-English model guides.
