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: 测试正确 ↓.
293/293
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） |
|---|---|---|---|---|---|---|---|
| #307 | Gemma 4 31B none | 1 | 0.6 | $0.031 | 9/23 | 7.28s | |
| #320 | Gemma 4 26B A4B none | 2 | 0.5 | $0.031 | 9/23 | 10.5s | |
| #324 | Qwen3.8 27B none | Qwen | 1 | 0.5 | ~$0.010 | 9/23 | 5.05s |
| #337 | Inkling Small low | Thinkingmachines | 2 | 0.5 | $0.055 | 9/23 | 1.99s |
| #384 | Ring 2.6 1t default | Inclusionai | 2 | 0.4 | $0.026 | 9/23 | 48.9s |
| #376 | GLM 5V Turbo none | Z.ai | 2 | 0.5 | $0.052 | 8/21 | 2.99s |
| #379 | Owl Alpha medium | Openrouter | 2 | 0.5 | $0.000 | 8/21 | 11.9s |
| #385 | Mimo V2 Omni default | Xiaomi | 1 | 0.4 | $0.021 | 8/21 | 2.44s |
| #422 | Grok Build 0.1 none | X AI | 2 | 0.3 | $0.547 | 7/19 | 28.7s |
| #253 | Qwen3.6 27B none | Qwen | 2 | 0.6 | $0.148 | 8/23 | 12.7s |
| #256 | Minimax M3.1 Flash medium | Minimax | 3 | 0.6 | $0.000 | 8/23 | 19.5s |
| #264 | Trinity Large Thinking low | Arcee AI | 3 | 0.6 | $0.918 | 8/23 | 95.2s |
| #275 | Trinity Large Thinking medium | Arcee AI | 4 | 0.6 | $0.862 | 8/23 | 84.0s |
| #278 | Qwen3.5 Plus 2026-04-20 none | Qwen | 2 | 0.6 | $0.216 | 8/23 | 18.5s |
| #280 | Granite 4.2 8B medium | IBM Granite | 1 | 0.6 | $0.029 | 8/23 | 50.5s |