Decision models API

One account, one jv_live_ key and one prepaid balance call every decision model. Use the Unified Decision API to switch models with a single field, or each model's Native API for its own request/response. Jev's original /api/v1/decide is unchanged.

Authentication

Every paid endpoint takes the same bearer key. Get one on the pricing page.

Authorization: Bearer jv_live_your_key_here
Content-Type: application/json

Unified Decision API — recommended

POST /api/v1/decisions. Send a model plus the shared state + questions shape; the response schema is the same for every model, with each provider's untouched reply included as provider_data. Switch models by changing one field: "jev" · "clef" · "clef-flash" · "openai-decisions".

curl -X POST https://jevtypesafeai.com/api/v1/decisions \
  -H "Authorization: Bearer jv_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "clef",
    "state": "Order R-208 was charged twice. Refund the extra payment.",
    "questions": {
      "route":  { "type": "choice", "instructions": "Which team handles this?",
                  "criteria": { "billing": "refunds", "account": "login", "other": "" } },
      "urgent": { "type": "noul", "instructions": "Is this urgent?" }
    }
  }'

Response — unified across models:

{
  "model": "clef",
  "provider": "Cloudflare",
  "answers": {
    "route":  { "type": "choice", "choice": "billing", "confidence": 0.96,
                "probabilities": { "billing": 0.96, "account": 0.02, "other": 0.02 } },
    "urgent": { "type": "noul", "noul": 0.63 }
  },
  "provider_data": { /* the provider's untouched raw response */ },
  "usage": { "input_tokens": 158, "cost_usd": 0.000095, "credits_remaining_usd": 4.99 }
}

Billing is also returned in headers: X-Decision-Model, X-Decision-Provider, X-Decision-Input-Tokens, X-Decision-Cost-USD, X-Decision-Credits-Remaining.

Native APIs

Prefer a provider's own request and response, byte-for-byte? Call its native endpoint. Same key and balance; billing rides in the X-Decision-* response headers. Full request shape and a copy-paste example are on each model's page.

Question types

• choice — pick one option; returns the choice, a probability per option, and confidence.
• score — an ordered scale; returns a weighted score.
• noul — a calibrated yes/no probability (OpenAI calls this predicate; the Unified API normalizes it to noul).

Pricing & errors

You're billed per 1M input tokens at each model's rate (see models); output tokens are free. Errors: 401 bad/missing key, 402 insufficient credits, 403 inactive account, 503 model unavailable or not yet priced.