Skip to content
SFAI
SFAI / PRACTICAL GUIDE

How to compare AI model API pricing

Compare the full workload, not just the headline input rate.

Start with your workload

Token-priced APIs commonly quote separate input and output rates. Estimate input tokens and output tokens independently, then multiply each by its matching rate. A model with inexpensive input can still cost more for a workload that generates long answers. The exchange calculator lets you compare these assumptions across offers.

Keep the quote conditions together

Compare currency, token unit, service tier, region and provider on the same offer. Do not pair one provider’s input price with another provider’s output price. Context thresholds, caching and batch discounts may change the applicable rate; inspect the source’s conditional pricing before deciding.

Read the observation date

A catalog quote is a published price observed at a point in time, not a binding offer. Source failures can leave an older observation in the catalog. Every profile identifies source links and observation dates so you can check the provider before deploying. Price history records observed changes rather than inventing a price for unobserved days.

Apply the method to current data

Browse current source-qualified model profiles · Review sources and publication limits