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: 失败次数 ↑.
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） |
|---|---|---|---|---|---|---|---|
| #2 | Gemini 3 Flash Preview medium | 1 | 8.6 | $0.667 | 2/3 | 84.4s | |
| #8 | Gemini 3.5 Flash low | 1 | 7.8 | $0.349 | 2/3 | 6.71s | |
| #10 | Gemini 3.1 Pro Preview medium | 1 | 7.9 | $1.054 | 2/3 | 40.2s | |
| #12 | GPT-5.5 medium | OpenAI | 1 | 8.8 | $3.679 | 2/3 | 59.8s |
| #14 | Qwen3.6 Max Preview medium | Qwen | 1 | 8.8 | $0.960 | 2/3 | 146.5s |
| #16 | Claude Opus 4.7 medium | Anthropic | 1 | 7.6 | $0.679 | 2/3 | 13.0s |
| #17 | GLM 5.2 medium | Z.ai | 1 | 8.2 | $0.129 | 2/3 | 41.0s |
| #20 | GPT-5.4 medium | OpenAI | 1 | 8.8 | $1.210 | 2/3 | 44.4s |
| #21 | Seed-2.0-Lite medium | Bytedance Seed | 1 | 8.0 | $0.175 | 2/3 | 156.7s |
| #22 | GPT-5.2 Chat none | OpenAI | 1 | 8.8 | $0.393 | 2/3 | 9.82s |
| #23 | Step 3.7 Flash medium | Stepfun | 1 | 8.8 | $0.376 | 2/3 | 27.4s |
| #26 | Grok 4.5 medium | X AI | 1 | 7.6 | $1.696 | 2/3 | 155.7s |
| #27 | DeepSeek V4 Flash high | DeepSeek | 1 | 7.8 | $0.027 | 2/3 | 50.6s |
| #28 | Gemini 2.5 Flash medium | 1 | 7.8 | $0.379 | 2/3 | 41.0s | |
| #29 | GPT-5.6 Terra high | OpenAI | 1 | 7.6 | $0.852 | 2/3 | 9.14s |