HyperAI
Command Palette
Search for a command to run...
Humanity's Last Exam
数学、科学、人文科学にわたる2,500の専門家検証済み問題からなるフロンティアレベルのベンチマーク。最終的な閉形式の学術評価として設計されている。
- モデル
- 267 現在、 684 合計
- 中央値
- 12.0%
- 範囲
- 1.1%–74.3%
| ランク | 比較する | モデル | スコア |
|---|---|---|---|
| 101 | 21.1% | ||
| 102 |  GPT-6 Luna (Low) OpenAI | 20.3% | |
| 103 | o3 OpenAI | 20.1% | |
| 104 | 19.9% | ||
| 105 | 19.8% | ||
| 106 | gpt-oss-120b (High) OpenAI | 19.6% | |
| 107 | 19.3% | ||
| 108 |  Gemini 3.5 Flash-Lite Google | 18.8% | |
| 109 |  Quasar 438B (Max, Based on GLM-5.2) Multiverse Computing | 18.7% | |
| 110 |  K-EXAONE 2.0 LG AI Research | 18.6% | |
| 111 |  K2 Horizon 7B Institute of Foundation Models | 18.2% | |
| 112 |  JT-4.1 Flash 236B A21B (Non-reasoning) China Mobile | 17.8% | |
| 113 |  G9v3-39A5B AI9Stars | 17.5% | |
| 114 | HyperNova 60B 2605 (High, Based on gpt-oss-120b) Multiverse Computing | 16.6% | |
| 115 |  KAT Coder Pro V2 KwaiKAT | 16.1% | |
| 116 | 15.9% | ||
| 117 |  Trinity Large Thinking Arcee AI | 15.8% | |
| 118 | Gemma 4 12B (Reasoning) Google | 15.7% | |
| 119 |  Qwen3.5 9B (Reasoning) Alibaba | 14.9% | |
| 120 | Qwen3.5 Omni Plus Alibaba | 14.9% | |
| 121 |  Doubao Seed Code ByteDance Seed | 14.1% | |
| 122 | Qwen3.8 27B (Medium) Alibaba | 14.1% | |
| 123 | Qwen3.8 27B (Low) Alibaba | 14.0% | |
| 124 | 13.9% | ||
| 125 | K2 Horizon 3.7B Institute of Foundation Models | 13.9% | |
| 126 |  Mistral Medium 3.5 Mistral | 13.8% | |
| 127 | 13.3% | ||
| 128 |  INTELLECT-3 (Based on GLM-4.5-Air) Prime Intellect | 13.1% | |
| 129 | EXAONE 4.5 33B (Reasoning) LG AI Research | 12.9% | |
| 130 | 12.6% | ||
| 131 | Qwen3.8 27B (Non-reasoning) Alibaba | 12.1% | |
| 132 |  Cogito v2.1 (Reasoning) Deep Cogito | 12.0% | |
| 133 | 12.0% | ||
| 134 |  Command A+ Cohere | 12.0% | |
| 135 | 11.8% | ||
| 136 |  Mercury 2.5 Inception | 11.8% | |
| 137 |  Nova 2.0 Lite (High) Amazon | 11.6% | |
| 138 | 11.5% | ||
| 139 | 11.5% | ||
| 140 | EXAONE 4.0 32B (Reasoning) LG AI Research | 11.4% | |
| 141 | 11.4% | ||
| 142 | 11.2% | ||
| 143 | North Mini Code Cohere | 11.1% | |
| 144 |  Sarvam 105B (High) Sarvam | 11.0% | |
| 145 |  Falcon-H1R-7B TII UAE | 11.0% | |
| 146 | gpt-oss-20b (High) OpenAI | 11.0% | |
| 147 |  Hermes 4 - Llama-3.1 405B (Reasoning) Nous Research | 10.9% | |
| 148 |  Nanbeige4.1-3B Nanbeige | 10.9% | |
| 149 | DiffusionGemma 26B A4B Google | 10.8% | |
| 150 |  Step3 VL 10B StepFun | 10.8% |
