API Cost Calculator

How token pricing turns into an API budget

API billing often separates tokens sent to a model from tokens generated by the model. If a provider charges one rate for input tokens and another for output tokens, estimate each component independently and add any per-request fee. For prices quoted per 1,000 tokens, one request costs (input tokens ÷ 1,000 × input rate) + (output tokens ÷ 1,000 × output rate) + request fees. Do not mix per-million and per-thousand price units in the same calculation.

For example, 500 input tokens at $0.002 per thousand and 800 output tokens at $0.004 per thousand produce a $0.0042 token charge for one request before request fees. At 10,000 identical requests, that becomes about $42 before discounts, retries, ancillary features, or other billing items. A monthly forecast must also specify how many days of activity are expected, rather than silently treating every month as an identical traffic period.

Providers may price cached inputs, long context, multimodal data, tools, batching, or premium tiers differently. Published model pricing can change, and unsuccessful or repeated requests can alter actual consumption. Use current vendor pricing and production usage logs to validate the estimate; this calculator models the values you enter, not real-time prices.