模型排行
来源为 Artificial Analysis 当前公开榜单。综合能力仅展示非估算的 AA 智能指数，速度单位为输出 tokens/s；不同推理配置分别列示。单价来自 OpenRouter 的同名基础模型，完整任务成本可能随推理量变化。
左右滑动表格，查看评分、价格与操作 →
| 排名 | 模型 / 厂商 | 智能指数Artificial Analysis | 上下文tokens | 输入 / 输出USD / 百万 tokens | 评测来源 | 操作 |
|---|---|---|---|---|---|---|
| 51 | GPT-5.6 TerraOpenAI xhigh | 38 | 1M | $2 / $12含阶梯 / 时段定价 | ||
| 52 | MiMo-V2.6-FlashXiaomi | 38 | 1M | $0.14 / $0.28 | ||
| 53 | Claude Haiku 5.5Anthropic high | 38 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 54 | DeepSeek V4 Pro 0813DeepSeek max | 36 | 1M | $0.66 / $1.98 | ||
| 55 | Claude Sonnet 5.5Anthropic low with fallback | 36 | 1M | $2 / $10 | ||
| 56 | DeepSeek V4 Flash VisionDeepSeek max | 35 | 1M | — / — | ||
| 57 | GPT-6 LunaOpenAI xhigh | 35 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 58 | Claude Haiku 5.5Anthropic medium | 34 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 59 | GLM-5.3Z AI low | 34 | 1M | — / — | ||
| 60 | GPT-5.6 TerraOpenAI high | 34 | 1M | $2 / $12含阶梯 / 时段定价 | ||
| 61 | JT-4.1 Flash 236B A21BChina Mobile | 34 | 262K | — / — | ||
| 62 | Qwen3.8 27BAlibaba xhigh | 34 | 256K | $0.24 / $2.2 | ||
| 63 | Gemini 3.8 FlashGoogle low | 33 | 1M | $0.83 / $4.13 | ||
| 64 | GPT-6 LunaOpenAI high | 33 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 65 | K2 Horizon 375B A23BInstitute of Foundation Models | 31 | 524K | — / — | ||
| 66 | Beam 1.0Reflection | 30 | 1.05M | — / — | ||
| 67 | GPT-5.6 TerraOpenAI medium | 30 | 1M | $2 / $12含阶梯 / 时段定价 | ||
| 68 | Kimi K3Kimi low | 30 | 1.05M | — / — | ||
| 69 | GPT-6 LunaOpenAI medium | 30 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 70 | Gemini 3.1 Pro PreviewGoogle | 30 | 1M | $2 / $12含阶梯 / 时段定价 | ||
| 71 | Claude Haiku 5.5Anthropic low | 29 | 1M | $0.1 / $0.5含阶梯 / 时段定价 | ||
| 72 | MiniMax-M3MiniMax | 29 | 1M | $0.3 / $1.2 | ||
| 73 | Qwen3.8 27BAlibaba medium | 28 | 256K | $0.24 / $2.2 | ||
| 74 | GPT-5.6 TerraOpenAI low | 27 | 1M | $2 / $12含阶梯 / 时段定价 | ||
| 75 | Quasar 438BMultiverse Computing max | 27 | 1M | — / — |
没有可靠匹配的价格或规格时显示「—」，保留数据缺口。只比较 API 价格