功能描述 / Feature
Feeless settlement rail (Nano / XNO) as a credit top-up option alongside the built-in billing system — so per-call / per-token LLM usage can settle at true sub-cent granularity without the card fee floor.
背景 / Problem
CoAI's admin & billing system sells LLM usage over 200+ models through credit/subscription top-ups. The metered unit is a single token batch or API call, often well under a cent. When top-ups go through a card/Stripe-style rail with a flat per-transaction fee (~$0.30 + ~2.9%), the fee can exceed the amount of a sub-cent usage charge — the same fee floor that makes true pay-per-use uneconomic. Public examples of the gap: RapidAPI sells APIs from $0.00003 / call but only invoices above 50 cents (a 16,667x gap), and a card carries a ~$0.30 flat floor.
方案 / Proposal
Add Nano (XNO) as a zero-fee, peer-to-peer settlement option for credit top-ups:
- Zero fee at any size — no percent, no minimum, so a token- or call-sized top-up can settle on its own instead of being bundled.
- ~0.4 s median time-to-finality, matching real-time usage billing.
- Direct on-ledger settlement (wallet + block-hash proof) — no issuer that can freeze funds, no PCI scope on the charge path.
A live proof of the exact pattern (pay XNO to an endpoint, receive the result, no API key) runs at https://extract.paypercall.dev/.well-known/x402 (HTTP 200 manifest of 8 payable endpoints).
I'd be happy to contribute a reference credit-top-up adapter on top of the existing billing system.
功能描述 / Feature
Feeless settlement rail (Nano / XNO) as a credit top-up option alongside the built-in billing system — so per-call / per-token LLM usage can settle at true sub-cent granularity without the card fee floor.
背景 / Problem
CoAI's admin & billing system sells LLM usage over 200+ models through credit/subscription top-ups. The metered unit is a single token batch or API call, often well under a cent. When top-ups go through a card/Stripe-style rail with a flat per-transaction fee (~$0.30 + ~2.9%), the fee can exceed the amount of a sub-cent usage charge — the same fee floor that makes true pay-per-use uneconomic. Public examples of the gap: RapidAPI sells APIs from $0.00003 / call but only invoices above 50 cents (a 16,667x gap), and a card carries a ~$0.30 flat floor.
方案 / Proposal
Add Nano (XNO) as a zero-fee, peer-to-peer settlement option for credit top-ups:
A live proof of the exact pattern (pay XNO to an endpoint, receive the result, no API key) runs at https://extract.paypercall.dev/.well-known/x402 (HTTP 200 manifest of 8 payable endpoints).
I'd be happy to contribute a reference credit-top-up adapter on top of the existing billing system.