Agent economy metrics for October 2026, with month-over-month change
Metric
October 2026
MoM
x402 transactionsCoinbase Base coverage expanded and backfilled from 2026-09-16. Earlier frozen aggregates retain their historical scope; this is tracked-address activity, not an exhaustive payment census.
5,087,812
—
x402 settled volumeCoinbase Base coverage expanded and backfilled from 2026-09-16. Earlier frozen aggregates retain their historical scope; this is tracked-address activity, not an exhaustive payment census.
$274,220
—
New ERC-8004 registrations
1,112,435
+1562.2%
Virtuals ACP memos
4,173
-74%
Base agentic-ecosystem transactions
124,098
-63%
Olas transactions (weeks starting in month)Complete weeks starting in the month; not a calendar-month total.
243,481
—
⚠ Tempo MPP events: a full set of dated daily observations and a post-close source timestamp are not available; omitted, not zero.
⚠ x402 coverage changed between these periods; month-over-month growth is withheld because the scopes are not comparable.
⚠ ERC-8004 registrations are concentrated. Across the month, 93.7% of 1,112,435 registrations across all tracked chains carried agent-card URLs on a single domain (centralcity.ai); excluding it leaves 70,178. Read the month-over-month figure as one application onboarding accounts, not broad-based adoption.
⚠ Virtuals ACP memos: counted on ACP's two legacy Base contracts only. Activity on Virtuals' current ACP contract is tracked separately in jobs and is not part of this figure, so the change describes the legacy contracts, not all ACP activity.
How this report is built
Every figure comes from a preserved, versioned snapshot of dated observations. Daily totals require coverage of every day in the month; weekly totals require complete weeks. Month-over-month changes use a fixed prior edition and are shown only when the measurement scopes are comparable. Units are deliberately not blended: a payment settlement, a registry registration, a commerce memo, and a channel event are different acts, reported side by side.
Anything the pipeline could not fully measure for the month is omitted and flagged rather than estimated. The preserved observations behind every included number are public and free — recompute this edition from monthly-reports.json, or read the methodology.
Edition and corrections
Edition 1, frozen November 9, 2026. First fixed edition. Previous editions and source hashes are retained in the snapshot file.