{
 "bench_id": "b005-eibun",
 "slug": "eibun-2026-09-30",
 "title": "英語のプレスリリースを訳す（4AI・6課題・各3回）",
 "field": "英文・翻訳",
 "field_key": "eibun",
 "round": 1,
 "run_date": "2026-09-30",
 "timezone": "Asia/Tokyo",
 "version": "v1",
 "conductor": "AInformation編集部",
 "funder": "AInformation（自社負担）",
 "sponsor": null,
 "method": {
  "tasks_n": 6,
  "repeats": 3,
  "ais_n": 4,
  "trials_total": 72,
  "trials_valid": 72,
  "excluded": [],
  "new_chat_each_time": true,
  "order_rotated": true,
  "order_by_repeat": {
   "1": [
    "ChatGPT",
    "Gemini",
    "Claude",
    "Copilot"
   ],
   "2": [
    "Gemini",
    "Claude",
    "Copilot",
    "ChatGPT"
   ],
   "3": [
    "Claude",
    "Copilot",
    "ChatGPT",
    "Gemini"
   ]
  },
  "timeout_seconds": 300,
  "scoring": "機械採点のみ（AIに採点させていない）",
  "materials_generated_at": "2026-09-29",
  "materials_note": "架空の株式会社サンプル商事（Sample Shoji）の英語プレスリリース。型番・都市名・数字は作るときに決めてある",
  "counted_outcomes": [
   "scored",
   "refused",
   "timeout"
  ],
  "excluded_outcomes": [
   "limit",
   "our_error",
   "human_check"
  ],
  "answer_cleanup": "画面から答えを読み取るとき、AIによっては読み上げ用の1行や考えている途中の要約が答えと同じ枠に入る。採点の前に「答えの先頭にあって、あとでそっくり同じ行がもう一度出てくる」ところまでを落とす。4AIとも同じ処理をする。落とす前の答えは answers/ にそのまま置いてある",
  "csv_url": "/data/research/bench/eibun-2026-09-30/latest.csv"
 },
 "ais": [
  {
   "name": "ChatGPT",
   "plan": "ChatGPT Pro",
   "model_shown": "中程度",
   "checked_at": "2026-09-29 23:38"
  },
  {
   "name": "Gemini",
   "plan": "Google AI Pro",
   "model_shown": "Flash 拡張",
   "checked_at": "2026-09-29 23:38"
  },
  {
   "name": "Claude",
   "plan": "Claude Max",
   "model_shown": "Opus 4.8 高",
   "checked_at": "2026-09-29 23:38"
  },
  {
   "name": "Copilot",
   "plan": "Microsoft 365 Personal",
   "model_shown": null,
   "checked_at": "2026-09-29 23:38"
  }
 ],
 "summary": {
  "headline": "6課題を各3回ずつ投げたところ、正解率が一番高かったのはChatGPTで15/18試行の83.3%だった（2位はCopilotの55.6%）",
  "winner": "ChatGPT",
  "note": "1回の検証の結果。日や課題が変われば順位は変わる"
 },
 "scores": [
  {
   "ai": "ChatGPT",
   "correct": 15,
   "trials": 18,
   "accuracy": 0.8333,
   "accuracy_display": 83.3,
   "seconds_median": 18.7,
   "seconds_min": 14.0,
   "seconds_max": 32.8,
   "consistency": 1.0,
   "consistency_display": 100.0,
   "source_alive": null,
   "source_alive_display": null,
   "instruction_follow": 1.0,
   "instruction_follow_display": 100.0,
   "failures": {
    "refused": 0,
    "timeout": 0,
    "limit": 0,
    "our_error": 0,
    "human_check": 0
   }
  },
  {
   "ai": "Gemini",
   "correct": 3,
   "trials": 18,
   "accuracy": 0.1667,
   "accuracy_display": 16.7,
   "seconds_median": 39.4,
   "seconds_min": 25.2,
   "seconds_max": 56.5,
   "consistency": 1.0,
   "consistency_display": 100.0,
   "source_alive": null,
   "source_alive_display": null,
   "instruction_follow": 1.0,
   "instruction_follow_display": 100.0,
   "failures": {
    "refused": 12,
    "timeout": 0,
    "limit": 0,
    "our_error": 0,
    "human_check": 0
   }
  },
  {
   "ai": "Claude",
   "correct": 9,
   "trials": 18,
   "accuracy": 0.5,
   "accuracy_display": 50.0,
   "seconds_median": 23.1,
   "seconds_min": 14.0,
   "seconds_max": 50.9,
   "consistency": 0.8889,
   "consistency_display": 88.9,
   "source_alive": null,
   "source_alive_display": null,
   "instruction_follow": 1.0,
   "instruction_follow_display": 100.0,
   "failures": {
    "refused": 0,
    "timeout": 1,
    "limit": 0,
    "our_error": 0,
    "human_check": 0
   }
  },
  {
   "ai": "Copilot",
   "correct": 10,
   "trials": 18,
   "accuracy": 0.5556,
   "accuracy_display": 55.6,
   "seconds_median": 17.1,
   "seconds_min": 14.3,
   "seconds_max": 20.3,
   "consistency": 0.8333,
   "consistency_display": 83.3,
   "source_alive": null,
   "source_alive_display": null,
   "instruction_follow": 0.5556,
   "instruction_follow_display": 55.6,
   "failures": {
    "refused": 0,
    "timeout": 0,
    "limit": 0,
    "our_error": 0,
    "human_check": 0
   }
  }
 ],
 "tasks": [
  {
   "task_id": "t1",
   "level": "易",
   "prompt": "次の英語のプレスリリースから、3つを書き出してください。\n\n決まり\n・答えはちょうど3行。型番／生産する都市の名前／CEOの名前 の順で書く\n・原文の書き方のまま写す。説明は書かない\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "contains_all",
   "answer_key": "（機械採点 contains_all）型番＝SM-FSU693／都市名＝Fukuoka／CEOの姓＝Mizuno",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 14.0,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 14.1,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.2,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 25.5,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_01_t1.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "refused",
       "seconds": 251.4,
       "chars": 60,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 25.2,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 165,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_01_t1.png"
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 91,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 20.0,
       "chars": 57,
       "why": "3項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_claude_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.4,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 14.4,
       "chars": 31,
       "why": "3項目すべてあり",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 16.9,
       "chars": 95,
       "why": "3項目すべてあり",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t1_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  },
  {
   "task_id": "t2",
   "level": "易",
   "prompt": "次の英語のプレスリリースから、3つの数を書き出してください。\n\n決まり\n・答えはちょうど3行。前年度の出荷台数／希望小売価格（円）／容量（ml） の順で書く\n・数は半角で書く。説明は書かない\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "exact_number",
   "answer_key": "（機械採点 exact_number）出荷台数＝80500（誤差±0）／価格＝9140（誤差±0）／容量＝500（誤差±0）",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 14,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 14.1,
       "chars": 16,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 14,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 37.7,
       "chars": 16,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_02_t2.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "refused",
       "seconds": 47.1,
       "chars": 56,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "refused",
       "seconds": 28.5,
       "chars": 54,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 67,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_02_t2.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 14.0,
       "chars": 69,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 20.0,
       "chars": 137,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_claude_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 14,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 14,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 14.3,
       "chars": 14,
       "why": "数値が違う：出荷台数 ×（正解80500・台 の行が無い）／価格 ×（正解9140・円 の行が無い）／容量 ×（正解500・ml の行が無い）",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t2_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  },
  {
   "task_id": "t3",
   "level": "中",
   "prompt": "次の英語のプレスリリースを日本語に訳してください。\n\n決まり\n・型番・都市名・人名・数字は原文のまま落とさずに訳文へ入れる\n・訳文だけを出す。前置きと解説は書かない\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "contains_all",
   "answer_key": "（機械採点 contains_all）型番＝SM-FSU693／都市名＝Fukuoka／CEOの姓＝Mizuno／出荷台数＝80,500または80500／伸び率＝27.3／価格＝9,140または9140／容量＝500／保温12時間＝12／保冷24時間＝24",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.4,
       "chars": 633,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.4,
       "chars": 625,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 29.5,
       "chars": 618,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "refused",
       "seconds": 31.5,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_03_t3.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "refused",
       "seconds": 37.7,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 56.5,
       "chars": 26,
       "why": "9項目中0項目あり・足りない 型番／都市名／CEOの姓／出荷台数／伸び率／価格／容量／保温12時間／保冷24時間",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 50.9,
       "chars": 535,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_03_t3.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 29.2,
       "chars": 617,
       "why": "9項目中7項目あり・足りない 都市名／CEOの姓",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 26.1,
       "chars": 512,
       "why": "9項目中7項目あり・足りない 都市名／CEOの姓",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_claude_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 20.1,
       "chars": 572,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 20.1,
       "chars": 551,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 90,
       "why": "9項目中0項目あり・足りない 型番／都市名／CEOの姓／出荷台数／伸び率／価格／容量／保温12時間／保冷24時間",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t3_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  },
  {
   "task_id": "t4",
   "level": "中",
   "prompt": "次の英語のプレスリリースに出てくる数字を、日本語で書き出してください。\n\n決まり\n・答えはちょうど4行。出荷台数／売上の伸び率／希望小売価格／容量 の順で書く\n・数は半角で、原文の値をそのまま書く\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "contains_all",
   "answer_key": "（機械採点 contains_all）出荷台数＝80,500または80500／伸び率＝27.3／価格＝9,140または9140／容量＝500",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 14.2,
       "chars": 22,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 22,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 14.1,
       "chars": 22,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "refused",
       "seconds": 196.3,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_04_t4.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "refused",
       "seconds": 49.6,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "refused",
       "seconds": 77.8,
       "chars": 54,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 109,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_04_t4.png"
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 19.9,
       "chars": 107,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.1,
       "chars": 113,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_claude_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 25,
       "why": "4項目すべてあり",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 20.3,
       "chars": 25,
       "why": "4項目すべてあり",
       "instruction_follow": false,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.3,
       "chars": 47,
       "why": "4項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t4_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  },
  {
   "task_id": "t5",
   "level": "難",
   "prompt": "次の英語のプレスリリースを、日本語で300文字以内に要約してください。\n\n決まり\n・300文字以内\n・型番・都市名・CEOの名前・数字は落とさずに入れる\n・要約だけを出す。前置きと解説は書かない\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "contains_all",
   "answer_key": "（機械採点 contains_all）型番＝SM-FSU693／都市名＝Fukuoka／CEOの姓＝Mizuno／出荷台数＝80,500または80500／伸び率＝27.3／価格＝9,140または9140／容量＝500／保温12時間＝12／保冷24時間＝24",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 26.5,
       "chars": 268,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.4,
       "chars": 266,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 32.8,
       "chars": 270,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "refused",
       "seconds": 208.8,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_05_t5.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 46.9,
       "chars": 219,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "refused",
       "seconds": 81.4,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 26.2,
       "chars": 442,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_05_t5.png"
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 32.1,
       "chars": 449,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "timeout",
       "seconds": 301.8,
       "chars": 0,
       "why": "時間切れ（制限時間まで答えが止まらなかった）",
       "instruction_follow": null,
       "answer_url": null,
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 248,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": false,
       "outcome": "scored",
       "seconds": 16.9,
       "chars": 253,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 255,
       "why": "9項目中8項目あり・足りない 都市名",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t5_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  },
  {
   "task_id": "t6",
   "level": "難",
   "prompt": "次の英語のプレスリリースを読み、JSON だけを出力してください。\n\n決まり\n・出力は JSON だけ。前後に説明文を書かない\n・形は {\"title_ja\": \"日本語の見出し\", \"summary_ja\": \"日本語の要約\", \"facts\": [\"原文の数字と固有名詞\"]} にする\n・facts には型番・都市名・CEOの名前・出荷台数・伸び率・価格・容量・保温の時間・保冷の時間を入れる\n\n---- プレスリリース ここから ----\nFOR IMMEDIATE RELEASE\n\nSample Shoji Co., Ltd. to Launch SM-FSU693, a New Insulated Bottle Line\n\nFukuoka, Japan — Sample Shoji Co., Ltd. today announced SM-FSU693, a new 500 ml insulated bottle,\nshipping from March 2027. The suggested retail price is 9,140 yen, excluding tax.\n\nThe company shipped 80,500 units of its current bottle line in the previous fiscal year,\nand expects the new line to lift bottle category revenue by 27.3% in the first full year.\nHaruka Mizuno, Chief Executive Officer of Sample Shoji, said the SM-FSU693 keeps beverages hot for 12 hours\nand cold for 24 hours, and that the lid can be taken apart into three parts for washing.\n\nProduction will take place at the company's Fukuoka plant. A limited color edition will follow in\nSeptember 2027. Sample Shoji operates two directly managed stores and supplies about 1,300 retailers.\n\nNote: Sample Shoji Co., Ltd. is a fictional company created by the AInformation editorial team for\nAI testing. It does not exist.\n\nMedia contact: Sample Shoji Co., Ltd., Public Relations\n---- プレスリリース ここまで ----",
   "scorer": "contains_all",
   "answer_key": "（機械採点 contains_all）型番＝SM-FSU693／都市名＝Fukuoka／CEOの姓＝Mizuno／出荷台数＝80,500または80500／伸び率＝27.3／価格＝9,140または9140／容量＝500／保温12時間＝12／保冷24時間＝24",
   "materials": [
    "/data/research/bench/eibun-2026-09-30/materials/press_release_en.txt"
   ],
   "per_ai": [
    {
     "ai": "ChatGPT",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.2,
       "chars": 408,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_chatgpt_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 23.4,
       "chars": 401,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_chatgpt_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 20.2,
       "chars": 400,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_chatgpt_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Gemini",
     "trials": [
      {
       "n": 1,
       "ok": false,
       "outcome": "refused",
       "seconds": 50.2,
       "chars": 56,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_gemini_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/gemini_06_t6.png"
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 41.0,
       "chars": 339,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_gemini_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": false,
       "outcome": "refused",
       "seconds": 65.2,
       "chars": 35,
       "why": "断られた（ポリシー）。不正解として分母に入れる",
       "instruction_follow": null,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_gemini_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Claude",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 26.2,
       "chars": 774,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_claude_1.txt",
       "shot": "/data/research/bench/eibun-2026-09-30/shots/claude_06_t6.png"
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 26.2,
       "chars": 696,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_claude_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 44.8,
       "chars": 821,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_claude_3.txt",
       "shot": null
      }
     ]
    },
    {
     "ai": "Copilot",
     "trials": [
      {
       "n": 1,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 354,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_copilot_1.txt",
       "shot": null
      },
      {
       "n": 2,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.0,
       "chars": 516,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_copilot_2.txt",
       "shot": null
      },
      {
       "n": 3,
       "ok": true,
       "outcome": "scored",
       "seconds": 17.1,
       "chars": 342,
       "why": "9項目すべてあり",
       "instruction_follow": true,
       "answer_url": "/data/research/bench/eibun-2026-09-30/answers/t6_copilot_3.txt",
       "shot": null
      }
     ]
    }
   ]
  }
 ],
 "limits": [
  "1日1回の測定である。日や課題が変われば順位は変わる",
  "プランは各AIの有料版（ChatGPT Pro／Google AI Pro／Claude Max／Microsoft 365 Personal）。無料版の結果ではない",
  "既定のモデル・既定のモードだけで測っている（Deep Research などは使っていない）",
  "課題は架空の株式会社サンプル商事の自作データ。実在の企業のデータでの結果ではない",
  "機械で採点できる課題だけを扱っている。文章のうまさ・読みやすさは測っていない",
  "ブラウザで人が使うのと同じ画面から測っている。APIの結果ではない",
  "画面に出ているモデル名を読み取れないAIがある回は、その欄を空（null）にしている。読み取れた回はそのまま載せている"
 ],
 "license": "CC BY 4.0",
 "attribution": "出典として AInformation調べ と、使ったページのURL・測定日を書いてください。",
 "page_url": "https://ainformation.jp/research/bench/eibun-2026-09-30"
}