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。
197/197
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） |
|---|---|---|---|---|---|---|---|
| #71 | Qwen3.7 Flash high | Qwen | 1 | 8.2 | $0.052 | 2/3 | 58.8s |
| #73 | Seed-2.0-Code low | Bytedance Seed | 1 | 7.0 | $0.767 | 1/3 | 109.7s |
| #76 | DeepSeek V4 Flash 0423 high | DeepSeek | 1 | 7.8 | $0.042 | 2/3 | 50.6s |
| #79 | Solar Pro 4 xhigh | Upstage | 1 | 5.8 | $0.050 | 1/3 | 292.2s |
| #80 | Nemotron 3 Ultra medium | NVIDIA | 1 | 8.4 | $0.780 | 2/3 | 26.5s |
| #81 | Claude Opus 5 none | Anthropic | 1 | 5.9 | $1.550 | 1/3 | 5.54s |
| #86 | Qwen3.5 Plus 2026-02-15 medium | Qwen | 1 | 6.6 | $0.459 | 1/3 | 180.7s |
| #88 | GPT-5.4 Mini medium | OpenAI | 1 | 8.4 | $0.786 | 2/3 | 57.9s |
| #89 | Grok Build 0.1 medium | X AI | 1 | 5.7 | $1.130 | 1/3 | 108.5s |
| #98 | Claude Sonnet 4.6 none | Anthropic | 1 | 5.5 | $0.661 | 1/3 | 5.19s |
| #100 | KAT-Coder-Pro V2.5 low | Kwaipilot | 1 | 7.8 | $0.400 | 2/3 | 24.9s |
| #104 | Claude Opus 4.8 none | Anthropic | 1 | 5.5 | $1.166 | 1/3 | 3.29s |
| #107 | Muse Glimmer 30B high | Meta | 1 | 8.8 | $0.535 | 2/3 | 96.0s |
| #110 | Muse Glimmer 30B medium | Meta | 1 | 8.4 | $0.280 | 2/3 | 33.7s |
| #111 | Qwen3.5-122B-A10B medium | Qwen | 1 | 6.0 | $1.047 | 1/3 | 114.5s |