AI API Cost Calculator
Choose a model and enter monthly token usage to compare token subtotals across official and third-party quotes.
Quotes use the model detail pages' currency, availability and freshness rules, with a recorded timestamp for each row. Per-request minimums, caching, context tiers and service terms can change the final bill.
Uncached token estimate in USD. Missing rates show โ; input rates never substitute for output rates. Cache writes, context tiers, per-request minimums, tools and taxes are excluded; unknown does not mean free. CNY quotes use a fixed reference of 1 USD = 7.25 CNY, not a live FX rate.
FAQ
What costs are included in the calculation?
Uncached input and output token subtotals only. Recorded cache-read rates are shown separately without assuming a hit rate. Cache writes, subscriptions, context tiers, tools, taxes and per-request minimums are excluded. Missing output prices are never replaced by input prices.
How often are prices updated?
Check each quote's recorded timestamp. Collection can be delayed or fail; relay quotes older than 48 hours are excluded here. Manually maintained official prices keep their recorded date, not an implied fresh verification date.
Why does the same model cost different amounts on different platforms?
Routes, pricing groups, billing conditions and terms differ. Neither official nor third-party providers are always cheapest. Check model identity, context limits and required features, then compare the same workload and validate a small test against an actual bill.
What if I can't find the model I'm looking for?
Search uses registered canonical model names. A missing model may lack a mapping or an eligible recent quote; this does not mean the model does not exist. Browse the API pricing hub for the wider inventory.