Most scraping APIs sell a plan. An agent that needs one page extracted, once, has a different option: a per-call x402 route with the price quoted before any payment.
Measured live on 2026-10-15
-
GET /api/web/extract-once?url=https://example.com→ HTTP 402 in 0.70 s, quoting $0.005 (5,000 atomic USDC, Base mainnet) per accepted structured extraction. -
GET /api/web/once-info→ HTTP 200, free: the product contract is inspectable before spend. - The Markdown read on the same rail quotes $0.001 per accepted page.
- The rail has settled 96 x402 payments on Base since 2026-09-02 (BaseScan read of the payTo wallet; method published on the canonical page).
Honest scope
One known public HTTPS URL in; structured JSON or clean Markdown out. Not a crawler, not browser automation, no authenticated sessions. For whole-site crawls and JS-rendered scraping, Firecrawl covers more ground. For a single page inside an agent loop — RAG ingestion, source capture, narrow extraction jobs — per-call pricing with no subscription is the fit.
Full evidence and claim receipts: the canonical link on this article.
Top comments (0)