Content Moderation Pre-Filter API — decisions before generation

Moderation is two decisions in a trench coat: a binary "is this violations-likely" gate, and an ordinal severity score for what slips past. Decision models emit both natively — a 0-1 probability you can threshold, and an interpolable score across 2-10 ordered levels — with output tokens free, so screening huge volumes costs input only.

Why decision models

• Yes/no gate: 0-1 calibrated probability for threshold rules
• Severity: ordinal score across 2-10 levels
• Output free — screening cost scales with input only
• Full audit trail with per-call usage logging

Run it as a pre-filter in front of your generative pipeline: reject or escalate before expensive models ever see the content, and log every decision with token-precise billing for audit.

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": "…"}}
    }
  }'