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 ↓.
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） |
|---|---|---|---|---|---|---|---|
| #92 | Solar Pro 4 xhigh | Upstage | 3 | 7.6 | $0.050 | 12/22 | 134.9s |
| #95 | GPT-5.4 Nano medium | OpenAI | 2 | 7.5 | $0.142 | 12/22 | 13.3s |
| #102 | GPT-5.4 Mini medium | OpenAI | 3 | 7.5 | $0.786 | 12/22 | 26.7s |
| #113 | Claude Sonnet 4.6 none | Anthropic | 1 | 7.3 | $0.661 | 12/22 | 7.67s |
| #118 | Qwen3.7 Plus none | Qwen | 1 | 7.3 | $0.106 | 12/22 | 12.1s |
| #121 | Kimi K2.6 medium | Moonshot AI | 2 | 7.2 | $1.247 | 12/22 | 115.9s |
| #131 | Grok 4.3 medium | X AI | 2 | 7.0 | $0.741 | 12/22 | 44.2s |
| #132 | GPT-5.6 Sol none | OpenAI | 1 | 7.0 | $0.201 | 12/22 | 2.17s |
| #145 | Qwen3.6 Flash medium | Qwen | 1 | 6.8 | $0.741 | 12/22 | 45.0s |
| #149 | Solar Pro 4 medium | Upstage | 2 | 6.8 | $0.047 | 12/22 | 127.6s |
| #151 | GLM 5.3 low | Z.ai | 1 | 6.8 | $0.257 | 12/22 | 13.6s |
| #162 | Gemini 3.5 Flash Lite low | 1 | 6.6 | $0.138 | 12/22 | 2.56s | |
| #167 | Qwen3.6 35B A3B medium | Qwen | 1 | 6.6 | $0.672 | 12/22 | 58.5s |
| #168 | Mercury 2.5 Preview high | Inception | 2 | 6.6 | $0.030 | 12/22 | 4.01s |
| #173 | Gemini 3.1 Flash Lite Preview none | 2 | 6.4 | $0.052 | 12/22 | 1.56s |