Compatible with TypeSafeAI's System One API, for decision models such as Jev. The official SDKs (typesafe-sdk, @typesafe-ai/sdk) work unchanged — set their base URL to this site.

This is not a chat endpoint. You send a state and a map of typed questions; you get back one structured answer per question — a choice, a score, or a yes/no probability — with no text to parse.

How the body is forwarded

Only model, state and questions reach the upstream. Any other top-level field is removed and listed in amux.droppedParams (and the x-amux-dropped-params header): the upstream rejects unknown top-level fields with an error that does not name them. Fields inside a question are forwarded as-is. Question contents are validated by the upstream; a failure comes back as 400 naming the field, e.g. questions.q.choice.criteria: Field required.

Limits

64K tokens per request, and 32K for state plus the single longest question. Requests clearly over either limit are rejected here with context_length_exceeded before reaching the upstream.

Billing

Input tokens only; output tokens are free. usage.output_tokens is still reported, and appears on your bill as a free line.

{
  "model": "typesafe/jev",
  "state": {
    "message": "Hi, my Stripe integration has been failing for 3 days. Please help ASAP."
  },
  "questions": {
    "department": {
      "type": "choice",
      "instructions": "Which team should handle `message`?",
      "criteria": {
        "billing": "Payment or subscription issues",
        "technical": "Bugs or integration problems",
        "sales": null
      }
    },
    "frustration": {
      "type": "score",
      "instructions": "How frustrated does the customer appear?",
      "criteria": [
        "Calm",
        "Frustrated but civil",
        "Very angry"
      ]
    },
    "is_urgent": {
      "type": "noul",
      "instructions": "`message` conveys urgency or time-sensitivity"
    }
  }
}