What do agent-buyable APIs charge per call?
Nobody tells a seller what to charge in the x402 economy. This is the cross-provider benchmark: for each capability, the median price a typical provider publishes, the percentile range around it, and the comps behind the number. Built from 28,477 distinct priced offers, rebuilt 2026-08-25.
Rate card categories
How these numbers are built
Prices come from what each endpoint publishes — and where a strict HTTP 402 check succeeded, from the price the endpoint actually quoted rather than its catalogue entry. Catalogues drift: about one in twenty-seven listed prices disagrees with the live quote, occasionally by more than 100×, which is why every quote is re-checked and re-decoded on each refresh.
The corpus is also heavily concentrated — a single bulk seller can register thousands of near-identical routes, so any raw per-route average just reports that one seller’s list price. Two corrections follow: templated route families are collapsed into distinct offers keyed by provider, title and price, and then each provider contributes the median of its own offers exactly once. What’s left is a neutral, cross-provider view of what a typical seller charges.
Prices are what providers publish in their 402 quote. They are not evidence of demand, and a confirmed quote is not a settled payment.
What the price does not tell you
A median is a starting point, not a purchase decision. These guides cover what a quote leaves out: whether the listed price matches what the endpoint really asks for, what a cheap entry price hides, and how to judge a service before an agent pays it.