AI BENCHY カテゴリ別失敗
ドメイン特化: 不正解
ドメイン特化
不正解
ドメイン特化 で 不正解 が起きやすいAIモデルを確認し、弱点を早く見つけられます。
| 順位 | モデル | 企業 | 不正解 件数 | カテゴリスコア | 正解テスト | 応答時間（平均） |
|---|---|---|---|---|---|---|
| #14 | Qwen3.6 Max Preview medium | Qwen | 3 | 2.9 | 0/3 | 95.9s |
| #18 | Qwen3.7 Plus medium | Qwen | 3 | 3.6 | 0/3 | 45.3s |
| #26 | Qwen3.6 Plus medium | Qwen | 3 | 2.9 | 0/3 | 29.6s |
| #29 | Qwen3.5-122B-A10B medium | Qwen | 3 | 2.9 | 0/3 | 63.4s |
| #31 | DeepSeek V4 Flash high | DeepSeek | 3 | 4.1 | 0/3 | 100.3s |
| #36 | Qwen3.5 Plus 2026-04-20 medium | Qwen | 3 | 2.9 | 0/3 | 53.1s |
| #39 | Qwen3.6 Flash medium | Qwen | 3 | 3.5 | 0/3 | 14.6s |
| #40 | Gemini 3.1 Flash Lite Preview medium | 3 | 3.0 | 0/3 | 4.21s | |
| #41 | Nemotron 3 Ultra 550b A55b medium | NVIDIA | 3 | 3.5 | 0/3 | 24.9s |
| #44 | Gemini 3.1 Flash Lite medium | 3 | 2.9 | 0/3 | 3.16s | |
| #45 | GPT-5.4 Mini medium | OpenAI | 3 | 4.1 | 0/3 | 65.3s |
| #53 | Gemini 3.1 Flash Lite high | 3 | 3.6 | 0/3 | 139.9s | |
| #63 | GPT-5.3 Chat none | OpenAI | 3 | 3.5 | 0/3 | 13.0s |
| #75 | Ring-2.6-1T medium | Inclusionai | 3 | 3.5 | 0/3 | 64.9s |
| #78 | Qwen3.6 27B medium | Qwen | 3 | 2.9 | 0/3 | 73.4s |