OpenAI の Decisions API とは？Jev・Clef との違いをまとめてみた

"input": [{
  "role": "user",
  "content": [
    // 指示：この写真の商品を検査してください
    {"type": "input_text", "text": "Inspect the product in this photo."},
    {"type": "input_image", "image_url": "data:image/png;base64,iVBORw0KGgo..."}
  ]
}]

{
  "model": "jev-latest",
  "state": "Checkout has been failing for every customer for the last hour.",
  "questions": {
    "urgent": {"type": "noul", "instructions": "Is this support request urgent?"}
  }
}

{
  "model": "jev-1.13.0",
  "answers": {
    "urgent": {
      "type": "noul",
      "noul": 0.97
    }
  },
  "usage": {
    "input_tokens": 284,
    "output_tokens": 20
  }
}

{
  "model": "gpt-6-luna",
  "input": "Checkout has been failing for every customer for the last hour.",
  "questions": [
    {"type": "predicate", "name": "urgent", "instructions": "Is this support request urgent?"}
  ]
}

{
  "model": "gpt-6-luna",
  "answers": [
    {
      "type": "predicate",
      "name": "urgent",
      "probability": 0.99
    }
  ],
  "usage": {
    "input_tokens": 166,
    "input_tokens_details": {
      "cached_tokens": 0,
      "cache_write_tokens": 0
    },
    "output_tokens": 0,
    "output_tokens_details": {
      "reasoning_tokens": 0
    },
    "total_tokens": 166
  }
}

curl https://api.openai.com/v1/decisions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6-luna",
    // 判断の材料：注文に対して2重に請求された
    "input": "I was charged twice for my order.",
    "questions": [{
      "type": "choice",
      "name": "department",
      // どの部署がこの苦情を担当すべきか
      "instructions": "Which department should handle this complaint?",
      "choices": [
        {"value": "billing", "description": "Payments, invoices, and refunds."}, // 支払い・請求・返金
        {"value": "technical", "description": "Problems using the product."}, // 製品利用に関する問題
        {"value": "shipping", "description": "Delivery and tracking."}, // 配送と追跡
        {"value": "other", "description": "Requests outside these categories."} // 上記以外の問い合わせ
      ]
    }]
  }'

{
  "model": "gpt-6-luna",
  "answers": [
    {
      "type": "choice",
      "name": "department",
      "choice": "billing",
      "probabilities": [
        {
          "value": "billing",
          "probability": 1.0
        },
        {
          "value": "technical",
          "probability": 0.0
        },
        {
          "value": "shipping",
          "probability": 0.0
        },
        {
          "value": "other",
          "probability": 0.0
        }
      ],
      "confidence": 1.0
    }
  ],
  "usage": {
    "input_tokens": 145,
    "input_tokens_details": {
      "cached_tokens": 0,
      "cache_write_tokens": 0
    },
    "output_tokens": 0,
    "output_tokens_details": {
      "reasoning_tokens": 0
    },
    "total_tokens": 145
  }
}

curl https://api.openai.com/v1/decisions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6-luna",
    // 判断の材料：Safari ではエクスポートに失敗するが Chrome では動作する
    "input": "Export fails in Safari but works in Chrome.",
    "questions": [{
      "type": "score",
      "name": "severity",
      // この問題はどのくらい深刻か
      "instructions": "How severe is this issue?",
      "levels": [
        {"label": "Cosmetic", "description": "Appearance only; no lost functionality."}, // 見た目だけで機能は失われていない
        {"label": "Workaround available", "description": "A task fails, but another way works."}, // 作業は失敗するが別の方法で回避できる
        {"label": "Fully blocked", "description": "A task fails with no workaround."} // 作業が失敗し回避策もない
      ]
    }]
  }'

{
  "model": "gpt-6-luna",
  "answers": [
    {
      "type": "score",
      "name": "severity",
      "score": 0.97,
      "probabilities": [
        {
          "value": 0,
          "label": "Cosmetic",
          "probability": 0.03
        },
        {
          "value": 1,
          "label": "Workaround available",
          "probability": 0.97
        },
        {
          "value": 2,
          "label": "Fully blocked",
          "probability": 0.0
        }
      ],
      "confidence": 0.96
    }
  ],
  "usage": {
    "input_tokens": 146,
    "input_tokens_details": {
      "cached_tokens": 0,
      "cache_write_tokens": 0
    },
    "output_tokens": 0,
    "output_tokens_details": {
      "reasoning_tokens": 0
    },
    "total_tokens": 146
  }
}