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: 总成本 ↓.
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） |
|---|---|---|---|---|---|---|---|
| #170 | Gemini 3.5 Flash Lite medium | 1.0 | 0.7 | $0.440 | 2/2 | 1.72s | |
| #396 | MiniMax M2.5 medium | Minimax | 0.8 | 0.4 | $0.440 | 1/2 | 621ms |
| #101 | Gemini 4 Argon low | 1.0 | 0.8 | $0.438 | 2/2 | 2.40s | |
| #140 | Nemotron 3.5 Super VL medium | NVIDIA | 1.0 | 0.8 | $0.437 | 2/2 | 4.34s |
| #118 | GPT-5.6 Terra low | OpenAI | 1.0 | 0.8 | $0.435 | 2/2 | 1.48s |
| #142 | Qwen3.5 Plus 2026-04-20 medium | Qwen | 1.0 | 0.8 | $0.435 | 2/2 | 20.2s |
| #248 | Beam 1.0 501B A23B low | Reflection | 0.8 | 0.6 | $0.432 | 1/2 | 1.14s |
| #216 | MiniMax M3 medium | Minimax | 1.0 | 0.7 | $0.422 | 2/2 | 6.14s |
| #100 | Qwen3.7 Plus medium | Qwen | 1.0 | 0.8 | $0.415 | 2/2 | 16.1s |
| #160 | Inkling Small high | Thinkingmachines | 1.0 | 0.7 | $0.398 | 2/2 | 3.94s |
| #301 | DeepSeek V4.1 Flash none | DeepSeek | 0.7 | 0.6 | $0.391 | 1/2 | 6.81s |
| #362 | Gemini 3 PRO Preview medium | 1.0 | 0.5 | $0.385 | 2/2 | 3.26s | |
| #300 | GPT-5.6 Terra none | OpenAI | 0.9 | 0.6 | $0.377 | 1/2 | 1.15s |
| #52 | Gemini 3.6 Flash low | 1.0 | 0.9 | $0.363 | 2/2 | 2.82s | |
| #36 | DeepSeek V4.1 Flash 1017 low | DeepSeek | 1.0 | 0.9 | $0.355 | 2/2 | 1.25s |