System One (TypeSafe Jev)

Jev is not a chat model. You send a state (text or JSON) plus a map of typed questions, and it returns typed answers with calibrated probabilities — a routing, scoring, or guardrail decision your code can branch on, with no text to parse. It is served at POST /v1/systemone on the same key as everything else.

Request

Three question types: choice picks one option from up to 255 you define, score rates against an ordered rubric of 2–10 levels, and noul returns the probability (0–1) that a yes/no statement holds. Questions are evaluated in parallel against one state; budgets are 64k tokens per request and 32k for the state plus the longest question. Text only, no streaming.

curl https://app.inferencehub.tech/v1/systemone \
  -H "Authorization: Bearer sk-prov-live-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jev-latest",
    "state": "Hi, my Stripe integration has been failing for 3 days and I am losing sales. Please help ASAP.",
    "questions": {
      "department": {
        "type": "choice",
        "instructions": "Which team should handle this",
        "criteria": {
          "billing": "Payment or subscription issues",
          "technical": "Bugs or integration problems",
          "sales": "Pricing or account questions"
        }
      },
      "frustration": {
        "type": "score",
        "instructions": "How frustrated the customer appears",
        "criteria": ["Calm, just stating facts", "Frustrated but civil", "Very angry, strong language"]
      },
      "is_urgent": { "type": "noul", "instructions": "The message conveys urgency" }
    }
  }'

Response

One answer per question, under the ids you chose. choice and score answers carry a confidence derived from the probability distribution; noul is the probability itself.

{
  "model": "typesafe-jev-1.13.0",
  "answers": {
    "department": { "type": "choice", "choice": "technical", "confidence": 0.78,
                    "probabilities": { "technical": 0.85, "billing": 0.15, "sales": 0.0 } },
    "frustration": { "type": "score", "score": 1.0, "confidence": 1.0,
                     "legend": { "0": "Calm, just stating facts", "1": "Frustrated but civil", "2": "Very angry, strong language" },
                     "probabilities": { "0": 0.0, "1": 1.0, "2": 0.0 } },
    "is_urgent": { "type": "noul", "noul": 1.0 }
  },
  "usage": { "input_tokens": 425, "output_tokens": 73 }
}

Models and pricing

jev-latest (also jev-1.13.0 / typesafe-jev-1.13.0) — currently Jev 1.13. Billed on input tokens only at $0.042 per 1M; output tokens are free. Errors come back in the OpenAI { error: { message, type } } envelope, for example at least one question is required (400) or model not found (404).

TypeSafe SDKs

The official typesafe-sdk (Python) and @typesafe-ai/sdk (Node) work unchanged — point their base URL at the gateway and use your InferenceHub key:

from typesafe_sdk import TypeSafeClient, Choice, Noul
client = TypeSafeClient(api_key="sk-prov-live-YOUR_KEY", base_url="https://app.inferencehub.tech/v1")
r = client.system_one(
    state=ticket_text,
    questions={"route": Choice(instructions="Which team?", criteria={"billing": "…", "technical": "…"}),
               "urgent": Noul(instructions="Is this urgent?")},
)
print(r.answers["route"].choice, r.answers["urgent"].noul)

Not for coding agents: Claude Code, Cursor, Codex and similar tools need a text-generating model and will not list Jev. Use it from application code for routing, classification, scoring, and confidence-gated guardrails.