tanod/x402-bazaar-endpoint-demand

x402 Bazaar: per-endpoint demand

One row per endpoint listed in the Coinbase CDP x402 Bazaar, with the 30-day usage numbers the Bazaar reports for it. The Bazaar is the discovery list AI agents use to find APIs they can pay for with USDC over HTTP 402.

Snapshot 2026-10-09: 34,062 endpoints on 2,158 hosts. 33,610 report at least 1 unique payer and 7,358 report 3 or more. Updated by Tanod. Last sync: 2026-10-10.

Files

File	What it holds
x402-bazaar-endpoint-demand.csv	Latest snapshot, one row per endpoint, sorted by unique_payers_30d descending
x402-bazaar-demand-history.csv	One aggregate row per snapshot date (1 date(s) so far)

Endpoint columns: snapshot_date, url, host, name, category, price_usd, network, pay_to, unique_payers_30d, calls_30d, last_called_at. History columns: date, endpoints, endpoints_with_payers, endpoints_3plus_payers, total_payers_30d_sum, total_calls_30d_sum.

What the usage fields mean

unique_payers_30d and calls_30d are the l30DaysUniquePayers and l30DaysTotalCalls values in the quality object of each resource in the CDP Bazaar discovery API: distinct paying wallets and total paid calls over the last 30 days. last_called_at is the Bazaar's lastCalledAt timestamp. These numbers are reported by Coinbase and have not been verified by Tanod. Self-purchases and test wallets are not filtered out, so small counts can be noise.

Method

Tanod crawls the public CDP Bazaar discovery listing daily. This table is every cdp_bazaar record present in the latest successful snapshot. price_usd is the first listed payment requirement, in USD. host is the URL hostname without www.. Nothing is excluded, including Tanod's own routes. In this snapshot, summed over all endpoints, unique_payers_30d adds up to 89,620 and calls_30d to 916,002. Top row: ax1.vc with 4,750 unique payers and 390,208 calls. last_called_at runs from 2026-09-09 to 2026-10-09 where present.

Limits

• The sum of unique_payers_30d counts the same wallet once per endpoint it paid, so it is not a count of distinct buyers.
• One snapshot per day; the history file holds only aggregates, not per-endpoint history.
• Counts can include one seller's own sweeps or test traffic. Check the per-host spread before drawing conclusions.
• Endpoints that left the Bazaar before the snapshot are not in the endpoint table.

Using it

from datasets import load_dataset
endpoints = load_dataset("tanod/x402-bazaar-endpoint-demand", "endpoints", split="train")
history = load_dataset("tanod/x402-bazaar-endpoint-demand", "history", split="train")

Links

• Analysis of the same data: What AI agents actually pay for over x402
• Live data files: tanod.dev/data
• Companion dataset, prices by category: tanod/x402-category-prices
• Query the index from an agent: tanod.dev/connect
• Per-category tables of which listings agents pay, built from this data: web search, PDF, OCR, speech to text, EVM chain reads, Solana reads, company enrichment, QR codes and barcodes, website screenshots, web page to Markdown, token price, weather, text embeddings, email verification, IP lookup, image processing, token risk checks, DNS lookup, domain availability, exchange rates, geocoding, IBAN and VAT validation, language detection, named entity extraction, OFAC sanctions screening, phishing URL check, phone number validation, public holidays, sentiment analysis, text summarization, WHOIS and RDAP

License

CC BY 4.0. Cite as: Tanod, "x402 Bazaar per-endpoint demand", https://huggingface.co/datasets/tanod/x402-bazaar-endpoint-demand.