AI BENCHY 分类失败
常识问答：答案错误
常识问答
答案错误
看看哪些 AI 模型在 常识问答 上最容易遇到 答案错误，更快找出薄弱点。
失败原因
| 排名 | 模型 | 公司 | 答案错误 次数 | 分类得分 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|
| #56 | Seed-2.0-Mini medium | Bytedance Seed | 1 | 3.0 | 0/1 | 56.8s |
| #57 | Qwen3.5-35B-A3B medium | Qwen | 1 | 3.0 | 0/1 | 177.4s |
| #58 | GPT-5.2 medium | OpenAI | 1 | 3.0 | 0/1 | 28.2s |
| #59 | DeepSeek V3.2 medium | DeepSeek | 1 | 3.0 | 0/1 | 84.0s |
| #60 | GPT-5.4 Mini medium | OpenAI | 1 | 3.0 | 0/1 | 30.1s |
| #61 | Claude Sonnet 4.6 none | Anthropic | 1 | 3.0 | 0/1 | 4.67s |
| #62 | MiMo-V2-Omni medium | Xiaomi | 1 | 3.0 | 0/1 | 234.2s |
| #64 | Gemma 4 31B none | 1 | 3.0 | 0/1 | 1.25s | |
| #65 | DeepSeek V4 Pro high | DeepSeek | 1 | 3.0 | 0/1 | 39.1s |
| #66 | Grok 4.20 medium | X AI | 1 | 3.0 | 0/1 | 63.5s |
| #67 | GPT-5 Mini medium | OpenAI | 1 | 3.0 | 0/1 | 9.99s |
| #68 | Gemini 3.1 Flash Lite minimal | 1 | 3.0 | 0/1 | 724ms | |
| #69 | Kimi K2.5 medium | Moonshot AI | 1 | 3.0 | 0/1 | 83.9s |
| #70 | Qwen3.6 27B medium | Qwen | 1 | 3.0 | 0/1 | 81.0s |
| #72 | GPT-5.5 none | OpenAI | 1 | 3.0 | 0/1 | 5.01s |