API Reference

ENDPOINTS,
ILLUMINATED

Every licensed engine resolves into a versioned REST endpoint. Submit inputs, receive decisions with full weighted traces, and retrieve immutable audit records — all over standard HTTPS.

Quick start

Run your first inference in under a minute.

curl -X POST https://api.gxport.ai/v1/engines/credit-risk-sentinel/infer \
  -H "Authorization: Bearer $GXPORT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": {
      "transaction_amount": 4200,
      "merchant_risk": 0.32,
      "device_trust": 0.88
    },
    "options": { "include_trace": true }
  }'

Authentication

Pass your API key as a Bearer token in the Authorization header. Keys are scoped per workspace and environment.

Rate limits

Default 600 requests/min per key. Batch endpoints count as one request. Contact us for dedicated throughput.

Latency

Median 38ms, p99 under 80ms for single-engine inference. Traces add <5ms overhead.

Reference

Core endpoints

POST/v1/engines/{engine_slug}/infer

Run inference

Submit an input payload and receive a decision with a full weighted trace.

Request body
{
  "inputs": {
    "transaction_amount": 4200,
    "merchant_risk": 0.32,
    "device_trust": 0.88
  },
  "options": { "include_trace": true }
}
Response
{
  "decision": "approve",
  "confidence": 0.94,
  "latency_ms": 41,
  "trace": [
    { "factor": "transaction_amount", "weight": 0.41, "contribution": 0.38 },
    { "factor": "merchant_risk",      "weight": 0.29, "contribution": -0.12 },
    { "factor": "device_trust",       "weight": 0.30, "contribution": 0.27 }
  ]
}
GET/v1/engines

List engines

Paginated list of all licensed cognitive engines available to your workspace.

Response
{
  "data": [
    { "id": "mdl_8f2a", "name": "Credit-Risk Sentinel", "domain": "Fintech" },
    { "id": "mdl_3c91", "name": "Threat Vector Detector", "domain": "Security" }
  ],
  "has_more": false
}
GET/v1/engines/{engine_slug}/trace/{decision_id}

Fetch decision trace

Retrieve the immutable decision trace for a previously returned decision.

Response
{
  "decision_id": "dec_7b41",
  "engine": "mdl_8f2a",
  "factors": [
    { "factor": "transaction_amount", "weight": 0.41, "contribution": 0.38 }
  ],
  "model_version": "v4.2.0",
  "created_at": "2026-06-30T11:04:22Z"
}
POST/v1/engines/{engine_slug}/batch

Batch inference

Submit up to 500 inputs in a single call. Returns one decision per input.

Request body
{
  "inputs": [
    { "transaction_amount": 4200, "merchant_risk": 0.32 },
    { "transaction_amount": 910,  "merchant_risk": 0.71 }
  ]
}
Response
{
  "results": [
    { "decision": "approve", "confidence": 0.94 },
    { "decision": "review", "confidence": 0.62 }
  ]
}

Generate a live endpoint for your engines

Select an environment in the Deployment Handshake to resolve licensed engines into a real API URL.

Open Deployment
Powered by Google TranslateTranslate