Scoring & Grading API — ordinal scores with calibrated confidence

Whenever a pipeline needs "which level is this?" — severity triage, essay rubrics, lead quality, SLA buckets — an ordinal score beats both free text (unparseable) and binary gates (too coarse). The score primitive returns an interpolable value across 2-10 ordered levels plus the full distribution.

Why decision models

• 2-10 ordered levels, interpolable score
• Full probability distribution over levels
• Calibrated confidence for auto/ human split
• Free output tokens

Because confidence comes calibrated, you can auto-grade the easy 80% and escalate the ambiguous 20% — the exact behavior grading pipelines need.

Runnable example

curl https://api.easyjev.online/v1/systemone \
  -H "Authorization: Bearer sk-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "typesafe/jev-latest",
    "state": {"user": {"plan": "pro"}},
    "questions": {
      "is_premium": {"type": "noul", "criteria": {"true": "…", "false": "…"}}
    }
  }'