▸ __BREADCRUMB__ usage
new: genuinely free AI models are live — try surp/free + see live budgets · token-gating prototype · vote on SRP

x402 gateway

402 — payment required. this gateway handles the rest.

An x402 gateway sits between a client and an API, verifying 402 payments and settling them on-chain. surp.ivc.lol is a production x402 LLM gateway: it paywalls AI inference per request in USDC on Base, so anyone with a wallet can call frontier models without an account.

what a gateway does

In the x402 flow, three parties cooperate:

  • Client — your app or agent. Sends the request, signs the USDC authorization.
  • Gateway — surp.ivc.lol. Presents the price (402), verifies the signature via a facilitator, forwards the request to the model.
  • Facilitator — submits the EIP-3009 transfer on-chain, sponsoring gas so the client only pays USDC.

our gateway adds dynamic pricing

Most x402 gateways charge a fixed price for a fixed resource. Ours is a marketplace router: the price and the model both come from the live Surplus Intelligence order book. The gateway picks the cheapest model in your requested class at that second, quotes that price, and on payment routes the request there. You never pay frontier prices for a model that's cheap on the open market.

why pay-per-request beats subscriptions for LLMs

  • No unused quota — you pay for exactly the tokens you use.
  • No credit card, no KYC — a wallet is enough.
  • No vendor lock-in — switch models per request, or let the gateway pick the cheapest.
  • Fully automated for agents — the handshake is protocol-level, not human-level.

the gateway as a product

If you build AI products, an x402 gateway can be your monetization layer: charge per request, settle in USDC, no billing infrastructure. The x402 reference implementation and our own docs cover the server side; connect your agent shows the client side.

related: what is x402? · x402 LLM API · pay-per-request LLM API · cheapest LLM API · docs