Jev

A decision engine — verdicts with confidence, not chat.

Jev answers structured questions about a situation you describe: yes/no checks, pick-one choices and graded scores, each returned with a confidence and the probability of every option. Built for gates and triage — does this evidence support shipping, which team should take this ticket, how risky is this change — where you want a calibrated verdict to act on, not prose to parse. Call POST /v1/systemone with a state object and a questions object. Best-effort capacity: a free upstream allowance backs it, so under heavy load the route can briefly refuse until its window resets.

Best for

Review gatesTicket triageModeration

Strengths

▸Calibrated confidence
▸Per-option probabilities
▸Tiny, cheap calls

Use Jev via AnyModel

One OpenAI-compatible endpoint — just set the model id:

curl https://anymodel.org/v1/chat/completions \
  -H "Authorization: Bearer $ANYMODEL_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "pg/jev-latest",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

One-line Codex install Get a key