Who pays for x402 on Base?

The study's six days show paid calls, dollars, paying wallets and wallets over 1,000 paid calls each. On every UTC day from September 28 through October 3 2026, 6 to 13 heavy wallets made 31.2% to 51.9% of calls and paid 1.8% to 5.3% of dollars. Wallets at 100 calls or fewer made 17.8% to 36.7% of calls and paid 77.1% to 95.3% of dollars. Dollars leave out two one-off transfers, $50,000 on September 28 and $34,700 on September 30, each to a wallet that received nothing else that day (see Corrections).

Stream coverage: direct payments submitted by any listed facilitator. Payments settled through a router contract were not in the study's stream: 0.8% to 12.5% of each day's payments, measured on the full census on October 4 2026 (12.5% on October 3, mostly Fluxa). Batch-settled payments, about 7%, are not in this stream.

Calls were concentrated before the heavy-wallet cutoff. On September 28, 10 wallets carried half of the day's calls. On October 2, 23 did. Across the six complete UTC days, the range was 10 to 23 wallets.

Two cohorts, 319 wallets paying 0x159b...4908 exactly $0.1000 and 214 wallets paying 0x5aae...482e exactly $0.2500, paid once a day on all six days and paid nothing else. Together they were 533 wallets, about 20% of a quiet day's payers. That reads as one operator's schedule, not broad demand.

On October 1 2026, 2,264 wallets that never paid before or after each paid one service, 0x8a12...0881, once. The payments ran from $0.5100 to $3, $4,770 in all. They were 43% of that day's payers and moved the payer-balanced median from about $0.0100 to $0.2500. The sampled $WAGE barely saw them because a one-call wallet enters the sample about 1 time in 16.

x402 exact payments are signed transfers submitted by a facilitator. The facilitator pays gas. A payer can pay a wallet it controls. That makes a loop of tiny self-controlled payments cheap, which is why Tickerz prints the heavy-wallet share beside $GIGS.

• Six complete UTC days, September 28 to October 3 2026, from the Goldsky stream in machine_payments on Base.
• The daily table is the study's six days, read from the same daily x402 table as Index Machina on October 4 2026 and fixed here; the live days continue there.
• A wallet is not a person. Heavy and cohort shares are floors for one operator's footprint.
• Nothing here says who ran the cohorts or the October 1 burst, or that they meant to move an index. They are patterns in public payments.
• The stream read direct payments submitted by any listed facilitator. It missed payments settled through a router contract (read since October 4 2026 15:07 UTC, and not in this study) and batch-settled payments.
• Prior work: Ling, Zhou, Wu and Wang, How Agentic Is Agentic Commerce? (July 14 2026), measured 280 days of x402 on Base and found 21.20% of settlements fictitious and 63.78% internal to linked clusters. This study is a six-day slice; Tickerz keeps the daily record.
• This is a measurement of payment rows. It is not advice.

Daily x402 table$GIGS$WAGEMethodology

October 4, 2026. Dollar shares now leave out two one-off transfers: $50,000 on September 28 (one payment) and $34,700 on September 30 (six payments from one wallet: $10,000, $8,000, $7,500, $6,700, $1,500 and $1,000). Each went to a wallet that received nothing else that day, so none is a payment for a call. The first version counted them, which put the heavy wallets' share of dollars at 0.3% to 5.3% and the light wallets' at 77.1% to 99.1%. Without them the ranges are 1.8% to 5.3% and 77.1% to 95.3%. The finding stands.

October 4, 2026. Coverage. The first version said the stream read three facilitators (Coinbase, UltravioletaDAO and OpenFacilitator). It read direct payments submitted by any listed facilitator, including Fluxa's; what it missed was payments settled through a router contract. Measured on the full census once Tickerz began reading routed payments, those were 0.8% to 12.5% of each day's payments (12.5% on October 3). No figure in the finding changes.

First published October 4, 2026.