The x402 landscape, measured.
HTTP 402 is the "payment required" status code being turned into a micropayment protocol — an agent gets a challenge, pays USDC on-chain, retries, and gets its data. We pull the full directory data and measure it end to end; this is the 23 September 2026 snapshot (the 19 September, 18 September and 5 September snapshots stay linked below for comparison). The numbers, how we got them, and the full dataset under a CC BY 4.0 license. Companion write-ups: the field guide to the x402 directories, and — what the directory's largest listing is actually made of, at wallet level — the x402 directory has a whale.
The headline, in one row
Quarter context: Q2 → now
The directory keeps a dated, corrected quarterly archive. Q2 2026 (April 1 – June 30) is the last closed quarter — a citable baseline, and the right thing to compare "now" against:
What the comparison actually says, scope-honestly: the directory grew 11× in about three months; uptime is down ~12 points (partly real, partly a different population and window); the rail is running at roughly half its Q2 monthly pace; and the measured per-service floor ($52.4k/30d) is only 8.2% of the measured facilitator rail — most settlement volume flows through endpoints the directory never lists. The widest lens of all is third-party on-chain measurement: in August 2026, x402 settlement contracts on EVM moved $2.59B, 88% of it through a single Arbitrum bridge contract (Bitquery data, independently verified by Pennyforge in September 2026; the whale article shows the audit method at directory level). "Directory" and "rail" are different animals; each number is correct inside its own scope. Full cross-check with methods and caveats: x402-q2-crosscheck-2026-09-23.json.
What a 402 service actually costs
Price = the minimum per-call price a service advertises. The market is a floor of dust: the median call costs a cent, and the mean ($0.41) is pulled up by a handful of expensive services. If you are pricing a 402 endpoint, you are pricing against this distribution, not against SaaS pricing.
| Price band (min price per call) | Share of 789 listed (771 priced) |
|---|---|
| Free ($0) | 2 services (0.3%) |
| Under $0.01 | 41.5% |
| $0.01 - $0.10 | 47.2% |
| $0.10 - $1 | 7.1% |
| $1 - $10 | 3.5% |
| Over $10 | 0.6% |
Two services are free. Nine in ten of the priced services sit under a dime.
New this month
308 services (39% of the listed set) were created in the last 30 days; 59 in the last 7. The churn isn't just new services — the directory also records constant schema changes (endpoints added or removed, prices changed) across the standing listings.
New services by category (30d): Data 102 · AI 68 · Verification 56 · Other 23 · Finance 21 · Content 19 · Blockchain 11 · Compute 8. Data + AI keeps taking the new supply.
Where the money actually goes
Read this before quoting a volume number. x402 settlement volume is measured on-chain per facilitator (USDC settled through each tracked facilitator address) — a measured floor: settlements outside the tracked facilitators aren't counted. Windows matter: the same ecosystem reports different totals for "today", "7 days", "30 days", and each window slides. And the flow has two faces: by settlement count, blockrun leads (~38% of measured settlements — 203k tiny tickets); by dollars, the top service is a checkout product (65.3% of the measured volume on ~$13.70 average tickets). The per-service view and the per-facilitator view are different slices of the same land rush (see the whale article and the census file). None of these numbers is wrong — they are different slices of the same land rush.
Facilitator axis (on-chain USDC, per window)
| Window | Settlement volume | Top facilitator (share) | Top-10 share | Facilitators with volume |
|---|---|---|---|---|
| Last 24 hours | $9,857 | Coinbase - 53.7% | 100% | 11 |
| Last 7 days | $167,971 | Coinbase - 49.7% | 99.8% | 18 |
| Last 30 days | $619,316 | Coinbase - 39.3% | 99.9% | 21 |
Top 5 facilitators by 30-day volume:
| # | Facilitator | 30d volume | 30d settlements |
|---|---|---|---|
| 1 | Coinbase | $243,635 | 1,712,181 |
| 2 | Meridian | $161,265 | 9,583 |
| 3 | Polygon Facilitator | $139,433 | 4,649,000 |
| 4 | FluxA | $54,390 | 272,890 |
| 5 | PayAI | $13,490 | 38,730 |
Cross-check: the directory's own measured traction (per-service, 578 measured services on Base + Solana) reports $52.4k / 497,769 settlements / 4,647 unique buyers over the same 30 days — a different, narrower measurement of the same floor (8.2% of measured flow). The 09-19 snapshot reported 1,400,370 per-service settlements; the provider's current settlement-identity dedup reports 497,769 for the same scope — volume and buyer series are stable across that change. Both axes agree: the flow is small, concentrated, and whale-driven.
Listed services by category
| Category | Services | Share |
|---|---|---|
| Data | 316 | 40% |
| AI | 144 | 18% |
| Verification | 95 | 12% |
| Finance | 92 | 12% |
| Blockchain | 46 | 6% |
| Other | 41 | 5% |
| Content | 37 | 5% |
| Compute | 16 | 2% |
| Infrastructure | 2 | 0% |
Data + AI is more than half the list. 257 services advertise more than one network (multi-chain 402 is common; Base is still the default).
Health: do they answer when they get paid?
Live status at capture time (of the 789 listed): 647 online, 58 degraded, 84 offline. Uptime rollups from the monitor: 83.7% (24h) and 85.3% (7d); the mean of the 24h-uptime field across the 789 listed services is 89.2%. Median response time 516ms; mean 726ms. The public list (789) lags the monitored universe (841) — pending-review and delisted services live in the difference.
"Verified" means delivery-verified — a probe that actually settled a payment and got a real response. Thirteen services hold it. Eight of the thirteen are one operator's product family (ForgeMesh — signals, fare intelligence, disruption intelligence, router, utility APIs, anomaly tracker, travel agent, ads), plus x402scan, KHOTEM, TariffMonkee, Losbeto and Scry Wallet Intelligence. The honest headline: five independent operators and one product family have proven delivery in a market of 789 listings.
Monitor at a glance
| Monitor metric (841 monitored) | Value |
|---|---|
| Endpoints tracked | 5,562 |
| Mainnet networks | 37 |
| Average uptime (24h / 7d rollup) | 83.7% / 85.3% |
| Average response time | 733 ms |
| Price: mean / median (monitored) | $0.40 / $0.01 |
Uptime and latency are the monitor's, not ours; we checked its method, not every probe.
The data (yours to use)
Everything on this page is a snapshot — captured 2026-09-23 07:12 UTC (service list) with aggregate stats computed at 06:41 UTC. If you cite it, cite it dated. Underlying data: x402-list.com (CC BY 4.0) — we fetched every page of its public /api/v1/services list plus /stats, /facilitators and the Q2 2026 state report, and computed the aggregates ourselves.
- Full 789-row dataset, 23 September (JSON, CC BY 4.0) — per service: name, endpoints, price, networks, uptime, latency, category, compliance grade, creation date. ~2.3 MB, one fetch.
- Computed summary, 23 September (JSON) — every aggregate on this page, machine-readable.
- Q2-vs-now cross-check, 23 September (JSON) — the quarterly archive, the live measurement and the third-party on-chain view, with methods, scopes and caveats.
- Previous dataset, 19 September (JSON, CC BY 4.0) — the 09-19 snapshot; the 09-19 → 09-23 volume change is window rollover (above), not a step change.
- Archived dataset, 18 September (JSON, CC BY 4.0) — the 09-18 snapshot.
- Archived 630-row snapshot, 5 September (JSON, CC BY 4.0) — the original snapshot; compare against it to see the weeks of growth and the volume drop.
- Archived summary, 5 September (JSON) — every aggregate of the original snapshot, machine-readable.
- Same dataset served by our x402 API (free, no paywall) — /landscape = summary, /landscape/full = full list.
- The field guide article — how each directory indexes services, how to get listed, what it costs.
- The x402 directory has a whale (18 September) — wallet-level measurement of the directory's largest listing: one EOA behind 97.6% of its 30-day volume, verified on-chain.
- The whale census file (JSON, CC BY 4.0) — every payer to that service's settlement wallet; re-verified 09-22 and re-scanned 09-23: the whale has been silent for 12 days (zero in or out across ~44k blocks) while the provider's flow continues as a small long tail — measured 24h on 09-23: $105.76 across 2,019 settlements.
Caveats: the public list (789) lags the monitored universe (841) — pending-review and delisted services live in the difference. Uptime and latency are the monitor's, not ours; we checked its method, not every probe. "Payment-ready" means the x402 challenge answers; it does not mean the response is worth the money. Measured by Pennyforge — a one-person studio running a $0.01 402 endpoint into this market, which is why we know these numbers.