Put them side by side.
Every row is a figure TOLL measured or a field a registry declared, and a row appears only where the same thing was measured the same way for every column. Where a column has no value the cell says which silence it is. Nothing here is a recommendation; the endpoint that reads best on a row is the one the number puts there.
| measure | cron Parse | serp |
|---|---|---|
| Declared price, per callmethodas the listing declares it; only comparable where the asset's decimals are on file | 0.01 USDClowest of these | 0.05 USDC |
| Calls, trailing 30 daysmethodthe registry's own counter, a trailing sum and not a rate | 3 | 2,041highest of these |
| Unique payers, trailing 30 daysmethoddistinct paying wallets over the same window, as the registry counts them | 1 | 12highest of these |
| Calls per payermethodcalls divided by payers, and nothing more: it says how concentrated the usage is, not how loyal anyone is | 3.0 | 170.1 |
| Verdict at the last sweepmethodread from the response body of a preflight call, never from its HTTP status | pass | pass |
| Sweeps passedmethoda fraction and never a percentage: between one and five sweeps have run, and a percentage over five observations would imply a precision the record does not have | 5 of 5 | 5 of 5 |
| Networks declaredmethodthe chains the listing says it will take payment on, not the ones it has been paid on | eip155:8453solana:5eykt4Us…qZKvdpmost of these | eip155:8453 |
| Position in its categorymethodby calls, among the endpoints in the same category that carry counters | #131 of 483 in Compute | #7 of 1,634 in Search |
| First observed by TOLLmethodobservation began 10 September 2026; a listing older than that shows that date | 2026-09-10 | 2026-09-10 |
| Paid tomethodthe payee on the primary accept, and how many listed endpoints name the same wallet | 0x548eFd…eEc5F6takes for 67 listed endpoints | 0x4df66B…3AcDeAtakes for 43 listed endpoints |
| Listed inmethodpresence in a registry at the last pull, which is not liveness of the service | Coinbase | Coinbase |
An uptime percentage is missing because between one and five daily sweeps have finished since observation began on 10 September 2026, and a percentage over five observations would imply a precision the record does not have; the sweeps-passed row shows the fraction instead. Growth over a window is missing because the counters are trailing thirty-day sums as the registry publishes them, so a difference between two of them is not a rate and TOLL will not present it as one. Both arrive when the record is long enough to carry them.