Inference for agents.Paid on demand.

Send OpenAI-compatible inference requests through a payment-aware endpoint, receive an x402 challenge, and settle accepted requests in RLUSD on the XRP Ledger—without an account or prepaid balance.

Inspect a 402 quote

11live model routes

402pay-per-request quotes

RLUSDsettlement on XRPL

$ Inference Router

clientPOST /v1/chat/completionsOUT

endpointHTTP 402 · payment requiredQUOTE

clientsign RLUSD payment · xrpl:0SIGN

clientreplay + PAYMENT-SIGNATURERETRY

xrplsettlement verifiedSETTLED

endpointHTTP 200 · response returnedIN

PAY PER REQUESTheurist/deepseek-v4-flash

VERIFIED ECOSYSTEM

Teams building with Heurist.

Official product links and brand assets for approved public references.

REQUEST PATH

CLIENT→HTTP 402 QUOTE→RLUSD ON XRPL→MODEL RESPONSE

01 · MODEL CATALOG

Live model routes. One payment-aware endpoint.

Reference prices are USD per one million tokens. Select a model to update the estimator and request examples below.

Catalog loaded from the router.https://router.heurist.xyz/x402/xrpl/v1/models

Availability and reference prices come directly from the router's public model catalog. The HTTP 402 challenge remains authoritative for the amount due.

02 · COST ILLUSTRATION

Size the request. Then inspect the real challenge.

This calculator applies only the displayed token reference rates. Endpoint minimums, margins, cache-read discounts, and current payment terms are intentionally excluded from the illustration.

REFERENCE SUBTOTAL(input tokens × input rate + output tokens × output rate) / 1M

MODELDeepSeek V4 Flash

ESTIMATE ONLY

Model route

Estimated input2,000 tokens

100log scale512K

Reserved output1,000 tokens

100log scale64K

ILLUSTRATIVE TOKEN SUBTOTALAt displayed reference rates

$0.00056≈ $0.56 per 1,000 requests

The current HTTP 402 challenge—not this calculator—defines the payment requirements.

03 · API EXAMPLES

Probe the quote. Replay with payment.

Start unpaid and inspect the returned 402 requirements. Sign an RLUSD payment only after the network, asset, issuer, payTo, and amount are verified, then replay the identical request.

CONFIGURED ENDPOINT: https://router.heurist.xyz/x402/xrpl
MODEL CATALOG: /v1/models
CHAT COMPLETIONS: /v1/chat/completions
PAYMENT HEADER: PAYMENT-SIGNATURE

curl -i -X POST "https://router.heurist.xyz/x402/xrpl/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "heurist/deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Contract probe"}],
    "max_completion_tokens": 128
  }'

Expected first response: HTTP 402 with a PAYMENT-REQUIRED header; no model response yet.

Pay per request. No account required.

OpenAI-compatible inference through x402, settled in RLUSD on the XRP Ledger.

Start with an unpaid probe Inspect model JSON