o1 Preview（オーワンプレビュー）

概要\n\no1 Preview(オーワン プレビュー)はOpenAI 2024年9月12日公開のo1 Previewで、Pro 業界初Mainstream Reasoning LLM Pro Famous + Pro Chain-of-Thought内部推論 + Pro Reinforcement Learning Reasoning + Pro AIME 83.3% / Codeforces 89%ile / GPQA Diamond 78% + Pro $15/$60 per 1M Token + Pro 業界初Mainstream Reasoning LLM Heritage継承代表機 + Pro Chain-of-Thought内部推論Heritage継承代表機 + 累計2024-2025年1年Heritage Pro Top独占代表機。o1歴史: 2023-Strawberry Project内部開発開始(Pro Q* Project Pro Famous噂) + 2024-09-12 o1 Preview公開Pro Famous Story(Pro 業界初Mainstream Reasoning LLM + Pro Chain-of-Thought内部推論 + ChatGPT Plus/Team対応) + 2024-09-12 o1-mini同時公開Pro Famous(Pro $3/$12 per 1M + Pro Coding特化) + 2024-12-05 Pro o1 Full公開Pro Famous(Pro 25%値下げ + Pro Multimodal Vision + Pro 200K Context) + 2024-12-20 Pro o3 announce Pro Famous Story(Pro AGI Benchmark ARC-AGI 87.5% Pro Famous類無し + Pro 推論Cost $20-3000/Task) + 2025-01-31 Pro o3-mini公開Pro Famous(Pro 高速 + Pro 安価 + ChatGPT Plus/Team) + 2025-04 Pro o3 Full + Pro o4-mini対応 + 累計2024-2025年1年Heritage継承。o1 Preview主要機能: (1)OpenAI主導(米国California州San Francisco + Sam Altman Pro CEO + Mark Chen Pro Sr Research VP)、(2)Pro 業界初Mainstream Reasoning LLM Pro Famous(Pro Mainstream LLM + Pro Reasoning統合 + Pro Strawberry Project由来)、(3)Pro Chain-of-Thought内部推論Pro Famous(Pro 内部Reasoning Token生成 + Pro 答え前思考 + Pro AIME/GPQA/Codeforces Top)、(4)Pro Reinforcement Learning Reasoning(Pro RL + CoT統合 + Pro 業界Pro Mainstream Reasoning出発点)、(5)Pro AIME 83.3% Pro Famous(Pro 米国数学オリンピック予選Top + GPT-4o 13.4%比較類無し)、(6)Pro Codeforces 89%ile Pro Famous(Pro 競技プログラミング業界Top独占)、(7)Pro GPQA Diamond 78% Pro Famous(Pro 大学院科学Top + Pro 人類専門家78%超越)、(8)Pro $15/$60 per 1M Token(Pro Top Tier Pricing)、(9)Pro o1-mini同時公開(Pro $3/$12 per 1M + Pro Coding特化 + Pro 5x安価)、(10)Pro 2024-12-05 o1 Full公開Pro Famous(Pro 25%値下げ + Pro Multimodal Vision + Pro 200K Context)、(11)Pro 2024-12-20 o3 announce Pro Famous Story(Pro AGI Benchmark ARC-AGI 87.5% + Pro 推論Cost $20-3000/Task)、(12)Pro 2025-01-31 o3-mini公開Pro Famous + Pro 2025-04 o3 Full + o4-mini、(13)Pro 業界初Mainstream Reasoning LLM Heritage継承代表機 + Pro Chain-of-Thought内部推論Heritage継承代表機 + 累計2024-2025年1年Heritage Pro Top独占代表機 + Pro業界History派 + Pro OpenAI派 + Pro Reasoning派 + Pro CoT派 + Pro AIME派 + Pro Codeforces派 + Pro Strawberry派 + Pro AGI派 真価発揮。o1 Preview vs 競合Reasoning LLM比較: o1 Preview(2024-09、本レコード、OpenAI + Pro 業界初Mainstream Reasoning + Pro $15/$60 + 1年Heritage)・o1-mini(2024-09、Pro $3/$12 + Coding特化)・o1 Full(2024-12、Pro Multimodal + 25%値下げ)・o3(2024-12 announce、Pro AGI Benchmark)・o3-mini(2025-01-31)・o4-mini(2025-04)・DeepSeek R1(2025-01-20、Pro 中国Open Source Reasoning + Pro 業界初Open Source Reasoning + Pro $0.55/$2.19 per 1M)・DeepSeek R1-Distill-Qwen 1.5B-70B(2025-01)・Qwen QwQ 32B Preview(2024-11-27、Pro Alibaba Open Source Reasoning + Pro 32B)・Gemini 2.0 Flash Thinking(2024-12-19、Google + Pro Multimodal Reasoning)・Claude 3.7 Sonnet Extended Thinking(2025-02-24、Anthropic + Pro Hybrid)・Grok 3 Think(2025-02-18、xAI + Pro Reasoning)、o1 Preview = Pro 業界初Mainstream Reasoning出発点 + Pro AIME 83% + Pro $15/$60 + 1年Heritage、o1-mini/Full = 同世代、o3/o3-mini/o4-mini = Pro後継、DeepSeek R1/Qwen QwQ/Gemini 2.0 Flash Thinking/Claude 3.7/Grok 3 = Pro主要競合。o1 Preview歴史的影響: (1)Pro 業界初Mainstream Reasoning LLM + 1年Heritage、(2)Pro Chain-of-Thought内部推論 + Pro Reasoning Token、(3)Pro Reinforcement Learning Reasoning統合、(4)Pro AIME 83.3% / Codeforces 89%ile / GPQA Diamond 78% Top、(5)Pro $15/$60 per 1M Token、(6)Pro o1-mini Pro $3/$12 + Pro Coding特化、(7)Pro o1 Full + Pro Multimodal Vision、(8)Pro o3 announce + Pro AGI Benchmark ARC-AGI 87.5%、(9)Pro o3-mini + Pro o4-mini後継、(10)Pro業界History派 + Pro OpenAI派 + Pro Reasoning派 + Pro CoT派 + Pro AGI派 真価発揮。Future: 2024-o1 Preview + 2025-o3/o4-mini + Pro Mainstream Reasoning LLM業界Top独占継続Heritage継続。\n\n## 主な特徴・仕組み\n\n- : 2024-09-12 OpenAI o1 Preview\n- : 2023-開発開始\n- \n- \n- \n- : GPT-4o 13.4%比較\n- \n- : 人類専門家超越\n- \n- : $3/$12 + Coding特化\n- : 25%値下げ + Multimodal Vision\n- : ARC-AGI 87.5%\n- \n- \n- \n\n## スペック比較表\n\n| Reasoning LLM | 公開年 | 開発元 | Pricing | Heritage |\n|---------------|--------|--------|---------|----------|\n| | 2024-09-12 | OpenAI | $15/$60 per 1M | Pro 業界初Mainstream出発点 |\n| o1-mini | 2024-09-12 | OpenAI | $3/$12 per 1M | Pro Coding特化 |\n| o1 Full | 2024-12-05 | OpenAI | $15/$60 per 1M | Pro Multimodal Vision |\n| o3 | 2024-12-20 announce | OpenAI | 推測高額 | Pro ARC-AGI 87.5% |\n| o3-mini | 2025-01-31 | OpenAI | API課金 | Pro 高速安価 |\n| o4-mini | 2025-04 | OpenAI | API課金 | Pro 後継 |\n| Qwen QwQ 32B | 2024-11-27 | Alibaba | Open Source | Pro 32B Open Reasoning |\n| Gemini 2.0 Flash Thinking | 2024-12-19 | Google | API課金 | Pro Multimodal Reasoning |\n| DeepSeek R1 | 2025-01-20 | DeepSeek | $0.55/$2.19 per 1M | Pro 業界初Open Source Reasoning |\n| Claude 3.7 Sonnet Extended Thinking | 2025-02-24 | Anthropic | $3/$15 per 1M | Pro Hybrid Reasoning |\n| Grok 3 Think | 2025-02-18 | xAI | API課金 | Pro xAI Reasoning |\n\n## 具体例・対応製品\n\n- \n- \n- \n- : ARC-AGI 87.5%\n- \n- \n- : OpenAI Pro CEO\n- : Pro Sr Research VP\n\n## 自作PCでの選び方・注意点\n\no1 Preview歴史Concept学習 + 現代Pro o系Workflow例: (A)現代Pro ChatGPT Pro $200/月構成: ChatGPT Pro + o3 + o4-mini + Pro 推論Top、(B)Pro代替¥0構成: DeepSeek R1 Open Source + Pro Local Inference + Pro 安価、(C)歴史Reasoning Heritage学習¥0構成: GPT-4 2023 → o1 Preview 2024-09 → o1 Full 2024-12 → o3 2024-12 → o3-mini 2025-01 → DeepSeek R1 2025-01 = 1.5年Pro Reasoning LLM Heritage学習Pro Reference。o1 Preview歴史 選択ポイント: (1)Pro 業界初Mainstream Reasoning LLM + 1年Heritage = 2024-09-12 OpenAI o1 Preview公開Pro Famous + Pro 業界初Mainstream Reasoning LLM Pro Famous + 累計2024-2025年1年Pro 業界初Mainstream Reasoning LLM業界出発点Heritage Pro Reference Heritage Pro Top独占 + Pro 業界初派 + Pro Reasoning派 + Pro 1年派 真価発揮、Pro 業界初派 + Pro Reasoning派 真価発揮、(2)Pro Chain-of-Thought内部推論 + Pro Reasoning Token = Pro Chain-of-Thought内部推論Pro Famous(Pro 内部Reasoning Token生成 + Pro 答え前思考 + Pro AIME/GPQA/Codeforces Top) + Pro Reinforcement Learning Reasoning統合 + 業界Pro Mainstream CoT/RL Reasoning業界出発点Heritage Pro Reference Heritage Pro Top独占 + Pro CoT派 + Pro RL派 + Pro Reasoning Token派 真価発揮、Pro CoT派 + Pro RL派 真価発揮、(3)Pro AIME 83% / Codeforces 89%ile / GPQA Diamond 78% Top = Pro AIME 83.3% Pro Famous(Pro 米国数学オリンピック予選Top + GPT-4o 13.4%比較類無し) + Pro Codeforces 89%ile Pro Famous(Pro 競技プログラミング業界Top独占) + Pro GPQA Diamond 78% Pro Famous(Pro 大学院科学Top + Pro 人類専門家78%超越) + 業界Pro Famous Reasoning Benchmark Top独占Heritage Pro Reference Heritage Pro Top独占 + Pro AIME派 + Pro Codeforces派 + Pro GPQA派 + Pro Top独占派 真価発揮、Pro Benchmark派 + Pro Top派 真価発揮、(4)Pro $15/$60 per 1M Token + Pro o1-mini $3/$12 = Pro $15/$60 per 1M Input/Output Token(Pro Top Tier Pricing) + Pro o1-mini同時公開(Pro $3/$12 per 1M + Pro Coding特化 + Pro 5x安価) + 業界Pro Mainstream Reasoning Pricing業界Heritage Pro Reference Heritage Pro Top独占 + Pro $15/$60派 + Pro $3/$12派 + Pro Coding特化派 真価発揮、Pro Pricing派 + Pro Coding派 真価発揮、(5)Pro o1 Full + Pro o3 + Pro o3-mini + Pro o4-mini Multi-Generation = Pro 2024-12-05 o1 Full公開Pro Famous(Pro 25%値下げ + Pro Multimodal Vision + Pro 200K Context) + Pro 2024-12-20 o3 announce Pro Famous Story(Pro AGI Benchmark ARC-AGI 87.5% Pro Famous類無し + Pro 推論Cost $20-3000/Task) + Pro 2025-01-31 o3-mini公開Pro Famous + Pro 2025-04 o3 Full + Pro o4-mini + 業界Pro Mainstream Reasoning LLM業界Top独占Heritage Pro Reference Heritage Pro Top独占 + Pro o1 Full派 + Pro o3派 + Pro AGI派 + Pro o3-mini派 + Pro o4-mini派 真価発揮、Pro Multi-Generation派 + Pro AGI派 真価発揮。\n\n## 関連用語との違い\n\n: OpenAI + 推測1.76T MoE + Pro Multimodal + Pro 8K-128K + 2年Heritage。o1 Preview(2024-09 + OpenAI + Pro 業界初Reasoning + Pro $15/$60 + 1年Heritage)後継 + 1.5年後継 + Pro 推測1.76T MoE → Pro Reasoning派生 + Pro Multimodal → Pro Text推論特化 + Pro 8K-128K → Pro 200K + 2年 vs 1年Heritage、GPT-4 = Pro 推測1.76T MoE + Pro Multimodal + Pro 8K-128K + Pro Microsoft Copilot基盤、o1 Preview = Pro 業界初Mainstream Reasoning + Pro Chain-of-Thought + Pro AIME 83% + Pro $15/$60。\n\n: DeepSeek + Pro 業界初Open Source Reasoning + Pro $0.55/$2.19 per 1M + Pro R1-Distill-Qwen 1.5B-70B + 数ヶ月Heritage。o1 Preview(2024-09 + OpenAI + Pro Closed Source + 1年Heritage)競合 + 4ヶ月後継 + Pro Closed Source → Pro Open Source(MIT License) + Pro $15/$60 → Pro $0.55/$2.19(27x安価) + Pro AIME 83% → Pro AIME 79.8%競合 + 1年 vs 数ヶ月Heritage + Pro 業界初Mainstream Reasoning vs Pro 業界初Open Source Reasoning、o1 Preview = Pro 業界初Mainstream Reasoning + Pro Closed Source + Pro $15/$60 + Pro AIME 83%、DeepSeek R1 = Pro 業界初Open Source Reasoning + Pro 中国DeepSeek + Pro $0.55/$2.19(27x安価) + Pro R1-Distill-Qwen系譜。\n\n## よくある質問(FAQ)\n\n\nA: 2024-09-12 OpenAI o1 Preview公開Pro Famous + Pro 業界初Mainstream Reasoning LLM Pro Famous + Pro Chain-of-Thought内部推論 + 累計2024-2025年1年Pro 業界初Mainstream Reasoning LLM業界出発点Heritage Pro Reference + 業界Pro Mainstream Reasoning LLM業界Top独占Heritage Pro Reference Heritage Pro Top独占 + 累計世界Pro Mainstream Reasoning LLM継承(o1-mini/o1 Full/o3/o3-mini/o4-mini/DeepSeek R1/Qwen QwQ/Gemini 2.0 Flash Thinking/Claude 3.7 Extended Thinking等継承)Pro Mainstream + 業界Pro Mainstream Reasoning LLM業界Top独占Heritage Pro Reference Heritage Pro Top独占。\n\n\nA: Pro AIME 83.3% Pro Famous(Pro 米国数学オリンピック予選Top + GPT-4o 13.4%比較類無し + Pro 6.2x精度向上Pro Famous) + Pro Codeforces 89%ile Pro Famous(Pro 競技プログラミング業界Top独占 + Pro 上位11%) + Pro GPQA Diamond 78% Pro Famous(Pro 大学院科学Top + Pro 人類専門家78%超越Pro Famous Heritage Pro Reference Heritage Pro Top独占) + 業界Pro Famous Reasoning Benchmark Top独占Heritage Pro Reference Heritage Pro Top独占 + 累計世界Pro Mainstream Reasoning Benchmark継承(o3 ARC-AGI 87.5%継承競争)Pro Mainstream + 業界Pro Mainstream Reasoning Benchmark業界Top独占Heritage Pro Reference Heritage Pro Top独占。\n\n\nA: 2024-12-20 Pro o3 announce Pro Famous Story(Pro AGI Benchmark ARC-AGI 87.5% Pro Famous類無し + Pro 推論Cost $20-3000/Task + Pro AGI Reference類無し) + 業界Pro Famous AGI Benchmark Heritage Pro Reference Heritage Pro Top独占 + 累計世界Pro Mainstream AGI Benchmark継承 + Pro 2025-01-31 o3-mini公開Pro Famous(Pro 高速 + Pro 安価) + Pro 2025-04 o3 Full + Pro o4-mini対応 + 業界Pro Mainstream AGI Reasoning LLM業界Top独占Heritage Pro Reference Heritage Pro Top独占 + 累計世界Pro Mainstream AGI継承(DeepSeek R1 ARC-AGI 15.8%/Claude 3.7 Sonnet/Gemini 2.0等継承競争)Pro Mainstream。\n\n## まとめ\n\n- 2024-09 o1 Preview、OpenAI Pro Reasoning出発\n- Chain-of-Thought + RL Reasoning\n- AIME 83% + Codeforces 89%ile + GPQA 78%\n- $15/$60 per 1M + o1-mini $3/$12 Coding\n- 2024-12 o3 ARC-AGI 87.5% + 1年Heritage