{"version":"2026-10-19.observations.2","checkedAt":"2026-10-19","series":{"id":"aime-answers","group":"capability","title":"AIME2025：限定数学题的答案成绩","question":"在同一公开赛事题组上，哪些答案被判对？","unit":"答案准确率 %","denominator":"AIME2025固定赛事题；实际每模型重复次数未知","protocol":"官方逐模型AIME2025行；服务档位保留；不是跨赛事IRT aggregate","scope":"Oct14有限横截面；运行日未知，模型日单列；横截面按模型/档位编号，日期另见卡片","summary":"精选6个模型/档位的公布值介于11.67至100.00（答案准确率 %）。公开题目可能被后发布模型见过；准确率不说明所有推导步骤成立或形式证明通过。 精选包括同系列high/xhigh以核对档位差异，按模型编号展示；不是完整61项或最优模型推荐。","nextAction":"按M01先确定验收答案还是推导，保留独立复算和全部尝试。","rows":[{"id":"obs-0382","label":"GPT-5.2 (xhigh)","value":100,"rawValue":"100.0","period":null,"modelDate":"2025-12-11","modelVersion":"openai_gpt_52","n":null,"ci":{"plusMinus":0,"level":"来源±区间，具体水平按原页"},"cost":0.18,"locator":"https://matharena.ai/models/openai_gpt_52；AIME2025 row","notes":"原页平均输出token=12811.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null},{"id":"obs-0392","label":"GPT-5.2 (high)","value":100,"rawValue":"100.0","period":null,"modelDate":"2025-12-11","modelVersion":"openai_gpt_52_high","n":null,"ci":{"plusMinus":0,"level":"官方页±区间，定义/置信水平本记录未取得"},"cost":0.11,"locator":"https://matharena.ai/models/openai_gpt_52_high；AIME2025 row","notes":"原页平均输出token=7758.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null},{"id":"obs-0383","label":"GPT-5 (high)","value":95,"rawValue":"95.0","period":null,"modelDate":"2025-08-07","modelVersion":"openai_gpt_5","n":null,"ci":{"plusMinus":3.9,"level":"来源±区间，具体水平按原页"},"cost":0.14,"locator":"https://matharena.ai/models/openai_gpt_5；AIME2025 row","notes":"原页平均输出token=13475.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null},{"id":"obs-0384","label":"o4-mini (high)","value":91.67,"rawValue":"91.67","period":null,"modelDate":"2025-04-16","modelVersion":"openai_o4_mini_high","n":null,"ci":{"plusMinus":4.95,"level":"来源±区间，具体水平按原页"},"cost":0.062,"locator":"https://matharena.ai/models/openai_o4_mini_high；AIME2025 row","notes":"原页平均输出token=13982.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null},{"id":"obs-0385","label":"DeepSeek-R1","value":70,"rawValue":"70.0","period":null,"modelDate":"2025-01-21","modelVersion":"deepseek_deepseek_r1","n":null,"ci":{"plusMinus":8.2,"level":"来源±区间，具体水平按原页"},"cost":0.025,"locator":"https://matharena.ai/models/deepseek_deepseek_r1；AIME2025 row","notes":"原页平均输出token=11211.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null},{"id":"obs-0386","label":"gpt-4o","value":11.67,"rawValue":"11.67","period":null,"modelDate":"2024-08-06","modelVersion":"openai_gpt_4o","n":null,"ci":{"plusMinus":5.74,"level":"来源±区间，具体水平按原页"},"cost":0.009,"locator":"https://matharena.ai/models/openai_gpt_4o；AIME2025 row","notes":"原页平均输出token=860.0；单行实际重复次数/运行日未知；模型晚于公开赛事可能见过题","valueStatus":"reported","sourceId":"C-AIME","actualCoverage":null,"runAt":null,"scorePublishedAt":null,"toolBudget":null}],"sourceIds":["C-AIME"],"xKind":"label","connect":false,"readingIds":["reading-measures"],"taskId":null,"boundaryTasks":["M01"],"actualCoverage":null,"evidenceStatus":"active"},"sources":[{"id":"C-AIME","title":"MathArena AIME2025固定赛事","url":"https://matharena.ai/","version":"2026-10-14官方模型详情页AIME2025行的有限摘要","locator":"逐模型详情页AIME2025 Accuracy/CI/Cost/Output Tokens","scope":"公开2025题组；模型/推理档位逐行保留","license":"网页数据许可未单独确认；仅少量公布事实摘要与原文链接，不镜像题目/产物","limit":"单行实际运行日与重复次数未知；赛事后模型可能见过公开题；不是研究级证明","publishedAt":null,"readAt":"2026-10-15","collectedAt":"2026-10-14","status":"active","nextReview":"2027-01-13","methods":["https://github.com/eth-sri/matharena"],"family":"C-AIME"}],"temporalRule":"观察期可画同来源趋势；模型发布日只展开当前横截面，未知运行/首次公开日不补。使用份额/企业采用/测试分不互换，不据此生成职业AI化比例。","actualCoverage":null}