# AIFIND モデルランキング — 総合

集計日時 (UTC): 2026-10-04T13:03:29.357Z

評価方法: 2026.10-observed-evidence-v17

スコアは相対指標です。正答率や将来の性能を表すものではありません。

| 順位 | モデル | スコア | 出典数 | 証拠のカバー率 |

| --- | --- | --- | --- | --- |

| 1 | Claude Opus 5.5 | 79.6 | 13 | 67.7% |

| 2 | Claude Fable 5 | 79.6 | 16 | 87.5% |

| 3 | Claude Fable 5.1 | 79.6 | 14 | 77.1% |

| 4 | Gemini 3.7 Flash | 77.4 | 15 | 84.4% |

| 5 | GPT-6 Astra | 77.0 | 16 | 87.5% |

| 6 | Gemini 3.8 Flash | 75.9 | 16 | 87.5% |

| 7 | GPT-6.1 Sol | 75.9 | 13 | 77.1% |

| 8 | Claude Opus 5 | 75.2 | 15 | 81.3% |

| 9 | Muse Spark 1.3 | 73.8 | 13 | 70.8% |

| 10 | GPT-5.6 Sol | 73.8 | 16 | 87.5% |

| 11 | GPT-6 Sol | 66.9 | 13 | 67.7% |

| 12 | Kimi K3 | 65.7 | 15 | 75.0% |

| 13 | Qwen3.8 Max | 65.6 | 14 | 81.3% |

| 14 | Muse Spark 1.2 | 65.4 | 11 | 71.9% |

| 15 | Gemini 3 Pro Preview | 64.6 | 6 | 34.4% |

| 16 | GPT-5.5 | 64.6 | 16 | 87.5% |

| 17 | Claude Opus 4.8 | 63.3 | 16 | 87.5% |

| 18 | Gemini 3.6 Flash | 62.1 | 15 | 84.4% |

| 19 | Gemini 3.5 Flash | 62.1 | 15 | 84.4% |

| 20 | Claude Sonnet 5.5 | 62.1 | 12 | 74.0% |

| 21 | Grok 4.6 | 60.1 | 14 | 81.3% |

| 22 | Grok 4.7 | 60.1 | 14 | 70.8% |

| 23 | GLM-5.3 | 59.2 | 15 | 75.0% |

| 24 | Claude Opus 4.7 | 58.0 | 15 | 83.3% |

| 25 | GPT-5.4 | 58.0 | 9 | 59.4% |

| 26 | GPT-5.6 Terra | 57.5 | 16 | 87.5% |

| 27 | Claude Opus 4.6 | 56.1 | 14 | 77.1% |

| 28 | DeepSeek V4.1 Flash | 56.1 | 8 | 49.0% |

| 29 | Gemini 3.1 Pro Preview | 56.1 | 16 | 87.5% |

| 30 | Muse Spark 1.1 | 54.1 | 11 | 71.9% |

評価方法と出典: /leaderboard/rules · /leaderboard/sources