This week's report → · All findings →
Measured directly from the Base USDC contract, not from self-reported dashboards.
We count transfers into each service's payTo address. Re-runnable by anyone.
This is a census of the whole Base catalogue: every USDC
transfer in a 24-hour window, joined against all
875 sellers in the public directory.
That is the entire Base catalogue, not the short list below: 8.0% of 875 listed x402 sellers took any payment at all in 24 hours, and 2.5% had more than one distinct payer. The money that did move was overwhelmingly bilateral: 48 single-payer sellers took $258.13 (83% of all value) at $1.16 a call, while the 22 sellers with a real multi-buyer business took $53.99 at $0.0737 a call.
Scope correction, measured 2026-09-10 after this census ran. The seller list
above comes from a pull of the public directory, so this measures directory-listed
sellers — not the Base x402 economy. We checked, and the two are very different populations.
One service that is not in the directory at all (claw402.ai, an
LLM gateway) received 329 USDC transfers in 67 minutes, about 7,100 a day from
10 payers. This entire census found 955 transfers across all 875 listed
sellers in a full day. A single unlisted seller out-transacts the whole catalogue we measured.
Treat $312.12 as what reached listed sellers, and not as the size of the market.
Coverage: 797/864 chunks read (92.2%).
67 chunks never returned data. A chunk that returns nothing is
indistinguishable from a chunk with no activity, and a missing chunk can only ever add —
so every number above is a floor, not a point estimate, and the file
itself records isComplete: false. Seller addresses come from a full
directory pull; volume is every USDC Transfer whose recipient is one of
them, scanned in 50-block windows. A single payer cannot be distinguished
from a service paying itself or a treasury sweep, and the largest seller's
$244.07 over 17 settlements, from 1
payer, is high enough per call that self-dealing is
a real alternative explanation. This measures where value moved, not who profited.
The 23 services below are the ones we started with, and they carry detail the census does not: median and largest ticket per service. They are a small slice of the 875 above, which is exactly why the census exists. DefiLlama's ecosystem-wide tracker put x402 settlement near $16k a day in July 2026, down from a ~$970k peak in December 2025 — orders of magnitude above the figures here. Part of that gap is the catalogue boundary; part of it is that a transfer into a listed seller is not the only way x402 value moves.
| Service | Settlements | Payers | Volume | Median |
|---|---|---|---|---|
| StableEnrich | 3,401 | 52 | $71.32 | $0.0100 |
| Agent402 | 178 | 16 | $1.12 | $0.0020 |
| toll402.dev | 74 | 1 | $0.3562 | $0.0010 |
| relay402.georgespring.worker | 3 | 1 | $0.1020 | $0.0010 |
| api.delx.ai | 23 | 2 | $0.0320 | $0.0010 |
| changelog-state-api.replit.a | 3 | 2 | $0.0300 | $0.0100 |
| rentcast.x402.paysponge.com | 2 | 1 | $0.0200 | $0.0100 |
| FiatDock | 1 | 1 | $0.0010 | $0.0010 |
| k2so.grok.me | 0 | 0 | $0.0000 | $0.0000 |
| market2000.xyz | 0 | 0 | $0.0000 | $0.0000 |
| base-pulse.annushka1190.work | 0 | 0 | $0.0000 | $0.0000 |
| radar-cnpj.com | 0 | 0 | $0.0000 | $0.0000 |
| simultaneously-provincial-pi | 0 | 0 | $0.0000 | $0.0000 |
| api.realedger.xyz | 0 | 0 | $0.0000 | $0.0000 |
| payforapi.com | 0 | 0 | $0.0000 | $0.0000 |
| chainray.online | 0 | 0 | $0.0000 | $0.0000 |
| chargers-borders-cedar-fig.t | 0 | 0 | $0.0000 | $0.0000 |
| earn-pulse.haiqingm58.worker | 0 | 0 | $0.0000 | $0.0000 |
| agentic-endpoints-testnet.ol | 0 | 0 | $0.0000 | $0.0000 |
| cloudmaxi0x.com | 0 | 0 | $0.0000 | $0.0000 |
| tar-experiencing-performance | 0 | 0 | $0.0000 | $0.0000 |
| sigtap-outreach-api.sigtap.w | 0 | 0 | $0.0000 | $0.0000 |
| agentdatapro.com | 0 | 0 | $0.0000 | $0.0000 |
Median ticket across the set: $0.0000. The buyers are countable on two hands, and the unit is half a cent.
Method. eth_getLogs against a public Base RPC, filtering
USDC Transfer events whose recipient is the service's payTo.
Self-reported counters are ignored entirely — we previously measured one catalogue whose
"sales" were 81% its own internal heartbeats.
Limitations. 24-hour window, so a single large
settlement would move the total. 23 services probed — payTo is absent from
most discovery documents, so the set is limited to services whose live 402 challenge we could read.
Base only. Inbound transfers only.
Excluded on purpose. Our own endpoint measured $25.28 over
4 settlements. That is operator funding and one internal test, not demand,
so it is not counted above. Stated because silently dropping it would be the same sin as the
counters we criticise. This row is labelled utunumus (us); it read
Ledger42 (us) in runs published before 10 Sep 2026, when the brand was unified.
It is matched on the wallet address as well as the label, because a rename is exactly how our
own volume leaked into the market total once already.