METER
A live report on the x402 agent economy
Base mainnet · press date pending · free to read, always
In this issue: Settlements· Who earns· What agents pay· Honest volume· The tape· Methodology· References
The headline number
--,---
x402 settlements pushed on Base by the tracked facilitator set, last 24 hours. Counted from chain state, in your browser, while you read this.
estimated volume, window
-- USDC est.
distinct buyers in sample
--
Fig. 1
Settlements over the window
Every x402 payment lands on Base as a facilitator transaction. The bars count them per interval: machine customers do not sleep, so neither does this chart.
-- the chart draws once the chain answers --
Settlements per interval, tracked facilitator set. Source: Base RPC, nonce deltas, measured live.
Fig. 2
Who is earning
Sellers ranked by receipts inside the live sample. A handful of services take most of the flow, which tells you how young this market still is.
-- waiting on sampled payments --
Top recipient addresses by sampled USDC received. Source: decoded transferWithAuthorization calldata from the sample scan.
Fig. 3
What agents pay per call
Ticket sizes in the sample. Most agentic purchases are fractions of a cent, the histogram shows where the money actually sits.
-- waiting on sampled payments --
Distribution of sampled payment amounts, log-ish bins. Median ticket: -- USDC.
Fig. 4
Honest volume, or test traffic
The narrative runs ahead of adoption. A blunt heuristic on the sample: payer-seller pairs that repeat in a short window look like tests or loops, diverse one-off pairs look like customers. Blunt, but disclosed.
-- waiting on sampled payments --
Sampled flow split by the repeat-pair heuristic. Definition in the methodology, hold us to it.
The tape
Latest sampled payments
| block | buyer | seller | tx | pattern | USDC |
|---|---|---|---|---|---|
| -- the tape prints when the sample lands -- | |||||
Methodology
How this page counts
This page has no backend and no stored dataset. Your browser asks public Base RPC endpoints and does the arithmetic locally.
Settlement counts. Coinbase runs a known set of x402 facilitator wallets. Each settlement is one transaction from one of them, so the wallet nonce is a payment odometer. We read eth_getTransactionCount at interval boundaries across the window and difference the readings. No sampling error, the chain keeps the ledger for us.
The sample. For everything money-shaped we scan recent blocks, keep facilitator transactions to the USDC contract, and decode transferWithAuthorization calldata: buyer, seller, amount, straight from the bytes. The tape, seller ranks, ticket sizes and the honesty split all come from this sample, and the page labels them as sampled, not total.
Estimated volume is the sampled average ticket multiplied by the counted settlements. It is an estimate and it says so on the label.
Real vs gamed. A payment joins the gamed bucket when its payer-seller pair appears more than three times inside the sample, or one seller takes over half of it. Crude, public, improvable: send a better rule.
The tracked set lives in config.js, currently three facilitator wallets confirmed on Basescan. When Coinbase spins up more, add them and the meter widens. The numbers here measure the agent economy that runs through x402 on Base, nothing broader, and we would rather undercount than perform.
References
Sources and the ticker
- x402 protocol documentation, Coinbase Developer Platform. docs.cdp.coinbase.com/x402
- Tracked facilitator wallets on Basescan: facilitator 1, facilitator 8, facilitator 10
- $METER contract: not printed yet, the press is warming up
$METER is a narrative index on the trend this page measures. It gates nothing: the data above is free for everyone, token or not. - USDC on Base: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913