Neuronto PaymentsAccept payments

x402 Endpoint Census

Of 3,291 hosts listing paid x402 endpoints in the public x402 discovery index, 3,084 (93.7%) answered at least one unpaid request with a valid payment challenge. Across the 16,179 endpoints probed, 93.8% returned a challenge an x402 client can pay, 0.4% served the response without asking for payment, and 0.3% could not be reached at all. Measured 2026-11-02, read only: nothing was paid or signed.

The census is weekly and the data is open under CC BY 4.0: JSON, CSV, one row per endpoint.

Headline numbers

MeasureValue
Resources in the index81,573
Unique resource URLs listed81,118
Hosts3,291
Hosts with at least one working paid endpoint3,084 (93.7%)
Hosts that answer, but with no valid challenge194
Hosts that could not be reached13
Endpoints probed16,179
Answered with any HTTP status99.7%
Answered with a valid x402 challenge93.8%
Same, weighted by how many resources each host lists98.4%
Median price of one call (USDC)$0.01
90th percentile price (USDC)$0.1
Median response time309.0 ms

What the endpoints answered

Every probed endpoint falls in exactly one class.

ClassEndpointsShareMeaning
Valid x402 challenge15,18393.8%answered 402 with a payment challenge an x402 client can read
402 without a usable challenge520.3%answered 402, with no readable challenge
Answered without payment660.4%answered 2xx and asked for nothing
Redirect490.3%answered 3xx (not followed)
Not found1190.7%answered 404 or 410
Method not allowed570.4%answered 405 to the method the listing declares
Wants other credentials400.2%answered 401 or 403
Other 4xx5333.3%answered another 4xx, such as 400, 422 or 429
Server error310.2%answered 5xx
Timeout90.1%no answer within the time limit
TLS error10%the certificate did not verify for the host name
Connection failed110.1%the connection was refused or reset
Name does not resolve80%the host name has no public address
Invalid URL200.1%the listed resource is not a usable http(s) URL

HTTP statuses among the endpoints that did not return a valid challenge: 400 (467), 404 (107), 200 (66), 405 (57), 402 (52), 406 (42), 302 (40), 403 (36), 422 (17), 503 (12), 410 (12), 530 (11).

The working endpoints

Of the 15,183 endpoints with a valid challenge:

FieldDistribution
x402 versionv2: 15,065, v1: 118
Challenge carried inheader: 15,072, body: 111
Schemes offeredexact: 15,152, upto: 185, batch-settlement: 51, wafermap-permit2: 19, exact_cosmos_authz: 17, settled-transfer: 16, borro-hold: 14, eip3009: 6
Networks offeredbase: 15,052, solana: 5,110, polygon: 3,376, arbitrum: 3,042, eip155:143: 1,134, xrpl:0: 1,131, eip155:480: 1,128, eip155:56: 881, algorand:wGHE2Pwdvd7S12BL5FaOP20EGYesN73k: 879, eip155:196: 864
Price of one callmin $0, median $0.01, 90th percentile $0.1, max $1,000,000.00
Challenge names a different resource URL than the one requested214

An endpoint usually offers several networks, so the network counts add up to more than the number of endpoints. Prices are counted only where the asset is USDC.

The index also reports usage. Of the probed endpoints taken from it, 553 (3.4%) show no paid call in the last 30 days.

History

One row per weekly run.

DateIndex sizeHostsWorking hostsProbedValid challengeAnsweredMedian price
2026-11-0281,5733,2913,08416,17993.8%99.7%$0.01

Largest listings

The 25 hosts listing the most resources, with what their sampled endpoints answered. The full per endpoint table is in the CSV.

HostListedProbedValid challengeMost common answer
market.datapackvibe.com48,6532020Valid x402 challenge
saylorinnovations.com1,6222019Valid x402 challenge
detect.halowerk.com1,3442020Valid x402 challenge
agent402.tools6232020Valid x402 challenge
agent-data-tools.revmesh2074.workers.dev5752020Valid x402 challenge
x402.orthogonal.com4092020Valid x402 challenge
api.x402node.dev3792020Valid x402 challenge
tanod.dev2782020Valid x402 challenge
twin.unykorn.org2762020Valid x402 challenge
api.vextorium.com2302020Valid x402 challenge
api.sitecheck-api.workers.dev2252020Valid x402 challenge
www.mahastrategies.com2152020Valid x402 challenge
buero.halowerk.com2092020Valid x402 challenge
api.myceliasignal.com2052020Valid x402 challenge
bot-bazaar.fly.dev1982020Valid x402 challenge
bess-x402.onrender.com1842020Valid x402 challenge
underground.substratesymposium.com1782020Valid x402 challenge
api.chainmarket-api.workers.dev1742020Valid x402 challenge
api.aiskills402.com1722020Valid x402 challenge
api.strale.io1652020Valid x402 challenge
x402.freeq.one1462020Valid x402 challenge
x402-niche-briefs.fly.dev1452020Valid x402 challenge
api.delx.ai1442020Valid x402 challenge
api.agentstools.dev1432020Valid x402 challenge
seo.halowerk.com1422020Valid x402 challenge

Method

  • Population. The full public x402 discovery index, every page, read at the start of the run, plus every resource in this facilitator's own catalogue. Duplicate URLs are counted once.
  • Sampling. Endpoints are grouped by host. A host listing 20 or fewer is probed in full; a larger one has 20 of its endpoints probed, chosen by a fixed hash of the URL so the same ones are probed every week. Host figures are therefore complete, and endpoint figures describe the probed endpoints. The weighted row scales each host's sampled shares by how many resources it lists; one host lists most of the index, so read that row with that in mind.
  • Request. One unpaid request per sampled resource, the listed method (GET or POST with an empty JSON body), no redirects followed, nothing paid or signed. Time limit 12 seconds. Addresses are checked before connecting, and private or reserved addresses are never contacted.
  • Classification. A challenge is valid when a 402 carries, in the PAYMENT-REQUIRED header (version 2) or the JSON body (version 1), an accepts list with at least one option naming a scheme, a network, a recipient and an amount.
  • Canaries. Each run must count this facilitator's own refunding endpoint as a valid challenge and a host under the reserved .invalid domain as unreachable, or the run is not published.

What this is not

It is not a ranking and not a judgement of any service. An endpoint can fail an unpaid probe for reasons its operator chose, such as requiring a request body the listing does not describe, and a listing can be retired faster than the index forgets it. The census reports what an agent following the index would find, measured the same way every week. Corrections to the method are welcome.

Data