# Claim Check by Agentropolis > Claim Check: send one factual claim, get supports / contradicts / unclear with a 0-1 confidence, a short explanation, cited source URLs and verbatim quotes that are machine-checked against the source text (no verified quote → unclear). Evidence from Wikipedia and GDELT news (recent 15-minute export files; DOC API when it answers). $0.02 standard, $0.08 deep (more sources, full news articles, larger judge model). Advisory, not a ruling. - POST https://seenly-kit-finder.vercel.app/api/claim with JSON {"claim":"The Eiffel Tower is about 330 metres tall."} → $0.02 USDC (x402, Base) - POST https://seenly-kit-finder.vercel.app/api/claim/deep (same body) → $0.08: more queries, full news article text, larger judge model - Unpaid calls get HTTP 402 (x402 v2 PAYMENT-REQUIRED). Settles only after a 2xx. Invalid input → 422 before payment; model/source outage → 503, not charged. - verdict: supports | contradicts | unclear (sources don't settle it). confidence 0-1. evidence[] has source URL, stance and a verbatim quote; quote_verified=true means we found it word-for-word in the source. A verdict without a verified quote is downgraded to unclear; numbers/dates in the claim must appear in a supporting source. - Sources: Wikipedia (CC BY-SA 4.0) and GDELT news (recent 15-minute export files; DOC API when it answers). Opinions and predictions return unclear with caveat not_checkable. - Sample output: https://seenly-kit-finder.vercel.app/samples/claim-sample.json · Kit manifest: https://seenly-kit-finder.vercel.app/claim/kit.json · OpenAPI: https://seenly-kit-finder.vercel.app/claim/openapi.json - Contact: hello@agentropolis.io