Quotes

How to price a run before you make it, using a search row's estimatedPrice and costPerSuccessUsd, or inspect's estimatedMaxCost.

There is no separate quote step. What a search row shows and what inspect shows before a run are the numbers you plan against.

On a search row

Every row carries estimatedPrice, priceBasis (per call, or per result) and costPerSuccessUsd (the price divided by the provider’s measured success rate, the better number for comparing providers that fail at different rates):

{
  "endpointId": "hunter-email-finder",
  "estimatedPrice": 0.0245,
  "priceBasis": "perCall",
  "costPerSuccessUsd": 0.0327
}

Before you run one endpoint: inspect

looot inspect hunter-email-finder

inspect { "endpointId": "hunter-email-finder", "detail": "run" }

inspect returns the exact price formula in endpoint.price and estimatedMaxCost, the most one run can hold. Running it needs available balance at least equal to estimatedMaxCost: check your balance first.

When an endpoint’s price depends on an input, estimatedMaxCost is the hold for the dearest option, and priceRange (minMicros, maxMicros, feeMicros, byInput, scalesWith) shows the spread. For example, linkup-search costs $0.005 with depth: "standard" and $0.05 with depth: "deep"; parallel-tasks-runs goes up to $2.40 depending on the processor; an OpenRouter chat call grows with max_tokens (scalesWith names that field). Pick the cheaper option when it does the job.

Signed-out search and inspect show the price block (which can list priceByInput) but not estimatedMaxCost or priceRange. On your own connected key a free call quotes estimatedMaxCost: 0 and has no priceRange. Check the signed-in quote before running an endpoint priced by an input.

Before you run a job

For job:<id> runs, price the job from its rows in a search answer, not from a single inspect call: inspect job:<id> is not supported yet. Search job:company.enrich (or filter by capability) and read costPerSuccessUsd on the row you expect to be picked.

Batch math

For a batch, multiply costPerSuccessUsd by the row count to estimate the total, and check your balance before you start. Tell the person running the batch the total when it is more than they would expect to spend.

For example, verifying 500 emails through job:people.email.verify at costPerSuccessUsd: 0.0033 estimates to about $1.65 for the batch; the charged total is the sum of what each run’s actualCost turns out to be, since a per-result price only charges for what came back. See Receipts.