Jev now available on MiniRouter

Jev, TypeSafe AI’s evaluation model, is now available on MiniRouter. Give it context and a set of questions. Get back probabilities, choices and scores your application can use directly.

Use it to classify support tickets, route requests, score an answer against a rubric, or check whether an agent completed a task. Multiple questions share the same context in a single request.

Make your first evaluation

Use your MiniRouter API key with typesafe-ai/jev on /v1/evaluate. Jev uses typed questions rather than chat messages.

curl --fail-with-body https://api.minirouter.sh/v1/evaluate \
  -H "Authorization: Bearer $MINIROUTER_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{
    "model": "typesafe-ai/jev",
    "state": "The subscription was charged twice.",
    "questions": {
      "billing": {
        "type": "boolean",
        "instructions": "Does this describe a billing problem?"
      }
    }
  }'

This example asks for the probability that a message describes a billing problem. You can also ask Jev to choose a category or return a score.

One key. Structured answers.

Jev works with the same MiniRouter key and credits you use for other models. Each settled request costs the verified provider inference cost plus 10%. The model page has current rates and availability.

Already using TypeSafe? The native format is available at /typesafe/v1/systemone. The evaluation guide covers JavaScript and Python setup, question types, payload limits and retry settings.

More about Jev: TypeSafe model documentation.