未遵循指令 失败
看看哪些 AI 模型最常遇到 未遵循指令，让你在选择前先发现稳定性风险。 排序方式: 分数 ↑.
215/215
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 | 未遵循指令 次数 | 分数 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #81 | Qwen3.8 27B medium | Qwen | 2 | 7.8 | ~$0.058 | 14/22 | 33.0s |
| #77 | Gemini 3.5 Flash Lite medium | 1 | 7.8 | $0.272 | 14/22 | 5.12s | |
| #76 | Qwen3.6 Plus medium | Qwen | 1 | 7.8 | $0.418 | 15/22 | 45.5s |
| #74 | Qwen3.8 27B low | Qwen | 1 | 7.9 | ~$0.071 | 15/22 | 39.1s |
| #73 | Step 3.7 Flash medium | Stepfun | 1 | 7.9 | $0.495 | 13/22 | 23.5s |
| #72 | GPT-5.2 Chat none | OpenAI | 1 | 7.9 | $0.594 | 13/22 | 7.73s |
| #68 | GLM 5.2 high | Z.ai | 1 | 8.0 | $1.188 | 14/22 | 69.9s |
| #67 | DeepSeek V4 Flash 0731 high | DeepSeek | 2 | 8.0 | $0.084 | 14/22 | 114.9s |
| #66 | DeepSeek V4 Flash 0731 low | DeepSeek | 1 | 8.0 | $0.044 | 13/22 | 77.6s |
| #65 | Muse Spark 1.1 high | Meta | 2 | 8.0 | $2.253 | 12/22 | 36.9s |
| #64 | Inkling high | Thinkingmachines | 1 | 8.0 | $1.046 | 15/22 | 62.6s |
| #63 | Muse Glimmer 30B xhigh | Meta | 1 | 8.1 | $0.471 | 13/22 | 95.6s |
| #62 | GPT-5 Mini medium | OpenAI | 3 | 8.1 | $0.241 | 12/22 | 27.6s |
| #60 | Claude Sonnet 5 medium | Anthropic | 1 | 8.2 | $0.922 | 15/22 | 12.5s |
| #59 | Muse Spark 1.3 low | Meta | 1 | 8.2 | $0.689 | 15/22 | 16.0s |