AI BENCHY 分类失败
编程：答案错误
编程
答案错误
看看哪些 AI 模型在 编程 上最容易遇到 答案错误，更快找出薄弱点。 排序方式: 测试正确 ↑.
134/134
筛选模型
没有模型匹配当前搜索和筛选条件。
| 排名 | 模型 | 公司 | 答案错误 次数 | 分类得分 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #60 | Qwen3.6 Flash medium | Qwen | 3 | 5.0 | $0.288 | 0/3 | 42.9s |
| #74 | GLM 5.2 none | Z.ai | 2 | 3.7 | $0.042 | 0/3 | 7.55s |
| #76 | Step 3.7 Flash high | Stepfun | 1 | 4.0 | $1.148 | 0/3 | 206.2s |
| #84 | Qwen3.5-Flash medium | Qwen | 2 | 3.7 | $0.080 | 0/3 | 58.9s |
| #87 | Mimo V2 Omni medium | Xiaomi | 1 | 3.3 | $0.683 | 0/3 | 183.9s |
| #116 | Qwen3.6 Max Preview none | Qwen | 3 | 3.8 | $0.075 | 0/3 | 3.12s |
| #118 | GLM 5 none | Z.ai | 3 | 4.0 | $0.027 | 0/3 | 5.12s |
| #123 | Qwen3.5 Plus 2026-02-15 none | Qwen | 3 | 4.3 | $0.016 | 0/3 | 2.05s |
| #124 | North Mini Code medium | Cohere | 3 | 4.5 | $0.000 | 0/3 | 320.4s |
| #131 | Mimo V2 Omni none | Xiaomi | 1 | 4.4 | $0.021 | 0/3 | 2.75s |
| #132 | Claude Sonnet 5 none | Anthropic | 3 | 4.6 | $0.287 | 0/3 | 3.67s |
| #134 | GLM 5.1 none | Z.ai | 3 | 3.9 | $0.057 | 0/3 | 4.96s |
| #135 | DeepSeek V4 Flash none | DeepSeek | 3 | 4.2 | $0.007 | 0/3 | 17.1s |
| #137 | MiMo-V2.5-Pro none | Xiaomi | 2 | 4.3 | $0.017 | 0/3 | 1.41s |
| #139 | Gemma 4 26B A4B none | 2 | 3.7 | $0.004 | 0/3 | 4.16s |