Where registration meets payment

Two layers that are usually discussed as one. This page joins them on the only thing they share — an address — and reports what survives. Every figure is scoped to the four pinned runs listed at the foot of the page, and every one of them is a lower bound.

measured 2026-07-30|354,858 agents|four chains

1 in 991

agents has ever been paid — 358 of 354,858. “Paid” means an external stablecoin transfer arriving after the agent was minted, at the wallet its own document declared.

1 in 10,437

has ever received a settlement through x402 — 34 agents, and 32 of them are on one chain. x402 is the payment protocol the ERC-8004 spec itself name-checks.

0.017%

of x402’s top-100 seller volume is agent-linked, tested against an index built by someone else, by a different method, that has no concept of ERC-8004.

Why x402, and how these figures were checked. “Paid” counts plain stablecoin transfers — that is the primary measure. On top of it, a payment is flagged x402-style when its transaction also carries an EIP-3009 authorization from the same token, which is how x402 settles on EVM chains. x402 is measured at the protocol level because it is the one payment protocol the ERC-8004 spec name-checks (a registration document may declare x402Support) and the one with an independent index to verify against; other rails — AP2, Solana’s — exist and are out of scope. The authorization signal is broader than x402 itself, so the 34 was checked against x402scan below: of the 20 receiving addresses it can see, 18 corroborate, 9 to the exact settlement count.

Per chain

lower bounds

Two stablecoins per chain and nothing else, so every value is a floor rather than a total. Symbols and decimals were read from each contract rather than assumed — BNB Chain’s USDC and USDT are 18 decimals, not 6.

chain	agents	ever paid	rate	ever x402	external value	from contracts
base	60,097	181	0.301%	32	$1,164,083	93.9%
bsc	244,208	128	0.052%	0	$1,639,492	86.5%
mainnet	40,806	31	0.076%	0	$203,121	68.1%
celo	9,747	18	0.185%	2	$2,813	50.8%
all four	354,858	358	0.101%	34	$3,009,509	—

Attestation does not predict payment. Celo attests at 79.5% and pays at 0.185% — a ratio of 430 to 1. Its attestation rate was three addresses writing feedback for one platform’s batch of agents; it was never a measure of commerce. No report should let one of these measures stand in for the other.

Checked against a second source

x402scan

The reverse direction, which is the stronger result: rather than asking whether our agents appear in an external index, we asked whether that index’s busiest sellers are registered agents.

top EVM sellers examined: 138
…that are a declared agentWallet: 0
…that own an agent, on any of the four chains: 3
x402scan's top 100 by volume: $49,617,581
our corroborated agent addresses: $8,596

The forward direction first: our 34 x402-settled agents resolve to 20 receiving addresses on chains x402scan indexes. 18 of the 20 corroborate — 9 to the exact settlement count. Both divergences trace to their index being facilitator-scoped. The hypothesis that would have overturned our number — double counting on our side — was tested first: every flagged transfer sits in its own transaction. No correction was forced.

Zero of the 138 is a declared agentWallet. The three that own an agent all sell on Base while owning an agent on a different chain — address reuse by an operator, not an agent being paid.

What this does not establish:a matching address confirms that a settlement occurred and who received it — never who initiated it or why. And a non-matching seller is not “not an agent”; it is not a registered ERC-8004 identity on the four chains we sweep. That is the finding: the registry is not where the payment activity is.

x402scan published no statement of limits of its own as of 2026-07-30, so this caveat is ours and is not attributed to them. Its index is facilitator-scoped where ours reads the chain directly, and it covers base, solana, polygon and optimism — not Celo, BNB Chain or Ethereum mainnet. www.x402scan.com, queried 2026-07-30.

Over time

one observation

This is the number worth watching, and there is exactly one of it. The chart below is deliberately not a trend — a line through a single point is a drawing, not a measurement.

Agents that have ever been paid, across all four chains. Updated per sweep; a rate that falls is as publishable as one that rises, and this project has no stake in which.

Provenance

reproducible

base: run cfbfcc01-fdaf-409f-9bed-abf706d865c7 · block 49,262,617
bsc: run f78c7891-e787-43f1-9748-61d5a361e9ff · block 112,874,357
mainnet: run 18a25593-9098-40fd-a0d2-75553c6ee31d · block 25,640,407
celo: run 7833fc49-a5b7-477b-99ce-946f650f0064 · block 73,448,013

The payments scan reads token transfer logs and EIP-3009 authorisations directly, so these figures are not produced by the census API and are not recomputed per page load. They are published numbers with a date on them. The full working, including the four corrections made before publication and the one caught during the cross-check, is in the report.

The full report →What each check measures →