Base 上 x402 市場數據快照呈現索引規模、價格與使用情況

Canonical version with full method and data: https://89-116-220-218.sslip.io/reports/state-of-x402-2026-10

If your agent pays for APIs over x402, what does the market it is paying into actually look like? We took a snapshot of the public CDP x402 Bazaar and are sharing the numbers, with the method and data behind each one. Thanks to the CDP team and every developer listing in the Bazaar; an open index is what makes this possible.

Snapshot: 2026-10-21 10:10 UTC (17:10 WIB) · Source: public CDP x402 Bazaar discovery API · Method: § Method · Data: aggregate_stats.csv, liveness_sample_200.csv

Every figure below carries its snapshot date, sample size (n) and method ID (M1–M4, defined in § Method).

1. Size of the index

Snapshot 2026-10-21 · n = 28,087 listings (full crawl) · Method M1 · data

Scope note: the CDP Bazaar is a Base-centric index, so the 99% Base share describes this index, not all of x402.

2. Multi-chain options

Snapshot 2026-10-21 · n = 28,087 · Method M1 · data

• 14,081 of 28,087 listings offer more than one payment option; 1,233 offer 13.
• Accept entries by network (top): Base 30,690 · Solana 13,557 · Polygon 7,658 · Arbitrum 6,105 · XRPL 3,143 · Monad 1,847 · Base Sepolia 1,563.
• Network identifiers appear in several forms (eip155:8453 and base — 230 entries; solana 50; solana:mainnet 11; shortened CAIP-2 strings). Client libraries that compare networks may want to normalise them.

3. Price on Base

Snapshot 2026-10-21 · n = 27,803 listings with a Base USDC option (cheapest Base option per listing) · Method M2 · data

Prices span six orders of magnitude, from $0.001 to $1,000. High prices can be entirely appropriate for the service; for buyer agents the practical point is to read the amount in the live challenge before signing.

4. Payout wallets (payTo) on Base

Snapshot 2026-10-21 · n = 27,803 listings with a Base option · Method M3 · data

• 1,706 distinct Base payTo addresses.
• 268 payTo addresses receive for listings on 2+ different hosts; 22 on 10+ hosts; the largest single payTo is shared by 101 hosts.
• The largest single payTo is attached to 1,630 listings.

Shared payout wallets are common and usually expected (platforms, marketplaces, operators running several domains). We publish counts only; no address is named, and wallet sharing is a fact to be aware of, not a signal about any operator.

5. Concentration of usage

Snapshot 2026-10-21 · n = 28,087 · Method M1 · data

• Median 30-day calls per listing: 2.
• Top 1% of listings: 84.1% of 30-day calls; top 10 listings: 55.4%.
• Last call within 24 h: 4,536 listings · within 7 days: 14,270.
• Hosts: median 2 listings per host; 1,117 hosts have exactly one listing; 47 hosts have 100+.

The market is early: most listings see light traffic, which is normal for a young ecosystem with many new services.

6. Live challenge sample

Snapshot 2026-10-21 · n = 200 randomly sampled GET listings (≈1.3% of GET listings) · Method M4 · data

Latency to first response: median 216 ms, p90 563 ms. With n = 200, percentages are indicative only.

The 3 differences have ordinary explanations as well (wallet rotation, an index entry not yet refreshed). We do not name them. The takeaway is simple: a directory listing is a snapshot, and the live 402 is what a buyer actually pays.

One mechanism that can help

Comparing the live challenge with the listing before signing is cheap, and anyone can do it. We built Pre-pay Check to package exactly these facts for agents: GET /x402/preflight?url=… ($0.01 USDC on Base) returns an Ed25519-signed JCS attestation with pass / fail / not_applicable checks (challenge validity, USDC on Base, payTo contract check, reachability and latency, payTo vs listing). It returns facts, not scores. Attestations can be verified offline with the published issuer key: https://89-116-220-218.sslip.io/reports/state-of-x402-2026-10/llms.txt

Method

• M1 — Full crawl. Public CDP x402 Bazaar discovery API, paged ?limit=1000&offset=n until exhausted, 2026-10-21 10:10 UTC. Counts are taken directly from listing metadata (accepts, x402Version, call counters, last-called timestamps).
• M2 — Base price. For each listing, the cheapest option with network eip155:8453 and asset USDC (0x8335…2913); price = amount / 10^6.
• M3 — payTo sharing. Lower-cased Base payTo addresses grouped by resource hostname. Only counts are published.
• M4 — Live sample. 200 GET listings drawn at random; one unpaid HTTPS GET each, UA x402-ecosystem-survey/0.1, 8 concurrent, 5 s timeout, no payment signatures sent. Live Base payTo compared with the listed payTo.
• Public files (https://89-116-220-218.sslip.io/reports/data): aggregate_stats.csv (aggregate counts only) and liveness_sample_200.csv (sample index, status, latency, header present, match true/false). Neither contains payTo addresses, hosts, URLs or raw listings. Raw crawl data is not published.
• Limits: one index, one point in time; call counters are as reported by the index; the live sample is small.

Questions or corrections are welcome in the comments. If a number looks off, tell us and we'll re-check it against the data.