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） |
|---|---|---|---|---|---|---|---|
| #130 | Solar Pro 4 high | Upstage | 3 | 7.1 | $0.046 | 11/22 | 120.6s |
| #128 | Muse Glimmer 30B low | Meta | 3 | 7.1 | $0.143 | 11/22 | 18.6s |
| #126 | Muse Glimmer 30B medium | Meta | 1 | 7.2 | $0.211 | 11/22 | 30.6s |
| #125 | Qwen3.7 Flash low | Qwen | 3 | 7.2 | $0.043 | 11/22 | 38.0s |
| #123 | Muse Glimmer 30B high | Meta | 3 | 7.2 | $0.397 | 10/22 | 66.7s |
| #121 | Kimi K2.6 medium | Moonshot AI | 2 | 7.2 | $1.247 | 12/22 | 115.9s |
| #120 | Claude Opus 4.8 none | Anthropic | 1 | 7.3 | $1.166 | 13/22 | 4.92s |
| #118 | Qwen3.7 Plus none | Qwen | 1 | 7.3 | $0.106 | 12/22 | 12.1s |
| #117 | Gemini 3.1 Flash Lite medium | 1 | 7.3 | $0.120 | 13/22 | 4.38s | |
| #116 | GLM 5.3 Flash high | Z.ai | 1 | 7.3 | $0.029 | 13/22 | 25.4s |
| #114 | GLM 5.3 high | Z.ai | 1 | 7.3 | $0.403 | 13/22 | 27.8s |
| #113 | Claude Sonnet 4.6 none | Anthropic | 1 | 7.3 | $0.661 | 12/22 | 7.67s |
| #110 | Qwen3.8 2.4T A95B high | Qwen | 2 | 7.4 | $2.357 | 14/22 | 120.6s |
| #107 | Kimi K2.7 Code medium | Moonshot AI | 1 | 7.4 | $0.687 | 11/22 | 80.1s |
| #105 | Gemini 3.1 Flash Lite Preview medium | 1 | 7.4 | $0.117 | 14/22 | 4.59s |