指令遵循 排名
看看哪些 AI 模型在 指令遵循 上表现最好，哪些更稳定，以及差距主要出现在哪里。 排序方式: 测试正确 ↓.
455/455
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 | 指令遵循 得分 | 分数 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #34 | Claude Sonnet 5.5 max | Anthropic | 0.9 | 0.9 | $8.557 | 1/2 | 6.24s |
| #40 | Muse Spark 1.3 high | Meta | 0.7 | 0.9 | $1.891 | 1/2 | 14.7s |
| #41 | Muse Wonder low | Meta | 0.7 | 0.9 | $2.637 | 1/2 | 2.08s |
| #44 | Muse Wonder minimal | Meta | 0.7 | 0.9 | $1.639 | 1/2 | 1.39s |
| #54 | Muse Spark Open low | Meta | 0.8 | 0.9 | $0.600 | 1/2 | 10.3s |
| #57 | Muse Spark 1.1 medium | Meta | 0.7 | 0.9 | $1.665 | 1/2 | 6.31s |
| #63 | Muse Spark Open medium | Meta | 0.7 | 0.9 | $1.260 | 1/2 | 18.4s |
| #70 | MiMo-V2.6-Pro medium | Xiaomi | 0.8 | 0.9 | $0.586 | 1/2 | 1.79s |
| #74 | Muse Spark 1.1 low | Meta | 0.7 | 0.9 | $0.885 | 1/2 | 5.42s |
| #84 | Nemotron 3.5 Super VL high | NVIDIA | 0.8 | 0.8 | $0.349 | 1/2 | 6.10s |
| #87 | Muse Glimmer 30B xhigh | Meta | 0.8 | 0.8 | $0.585 | 1/2 | 8.90s |
| #90 | Muse Spark 1.1 high | Meta | 0.6 | 0.8 | $2.570 | 1/2 | 7.81s |
| #109 | DeepSeek V4 Pro 0423 high | DeepSeek | 0.8 | 0.8 | $0.149 | 1/2 | 8.73s |
| #127 | Muse Wonder max | Meta | 0.8 | 0.8 | $4.766 | 1/2 | 13.2s |
| #134 | Qwen3.7 Plus none | Qwen | 0.6 | 0.8 | $0.178 | 1/2 | 929ms |