深度
比较DeepSeek-R1
deepseek/deepseek-r1
Classic open reasoning model for transparent math, coding, and deliberate problem solving
- 上下文窗口
- 13万
- 输入
- 文本
- 输出
- 文本
- 发布日期
- 2025-01-20
- 最大输出
- 3.3万
- 知识截止
- 2024-07
- 模型参数量
- 6845亿
- 激活参数量
- 370亿
- 权重
- 开放
- 许可证
- mit
- 能力
- 开放权重, 推理, 温度参数, 工具调用
输入
$0.574
USD / 1,000,000 token
输出
$2.294
USD / 1,000,000 token
表现
综合能力分
40.9/100
- 编程
- 38.1
- 推理
- 37.7
- 专业工作
- 31.0
- 知识与语言
- 39.1
| AA-LCRDeepSeek R1 (Jan '25) | 知识与语言 | Artificial Analysis | 2026-11-05 | |
| Terminal-Bench 4.0（AA 复跑）DeepSeek R1 (Jan '25) | 编程 | Artificial Analysis | 2026-11-05 | |
| AA-Briefcase | 专业工作 | Artificial Analysis | 2026-11-05 | |
| Humanity’s Last ExamDeepSeek R1 (Jan '25) | 推理 | Artificial Analysis | 2026-11-05 | |
| IFBench（Artificial Analysis）DeepSeek R1 (Jan '25) | 知识与语言 | Artificial Analysis | 2026-11-05 | |
| GDPval-AA | 专业工作 | Artificial Analysis | 2026-11-05 | |
| CritPtDeepSeek R1 (Jan '25) | 推理 | Artificial Analysis | 2026-11-05 | |
| BullshitBench v2 | 知识与语言 | BullshitBench | 2026-11-05 |
提供商
| 输入 | 输出 | 缓存写入 | |||||
|---|---|---|---|---|---|---|---|
| Alibaba (China) | deepseek-r1 | Custom | - | - | - | ||
| AnyAPI | deepseek/deepseek-r1 | Openai chat | - | - | - | - | - |
| Azure | deepseek-r1 | Openai responses | - | - | - | ||
| Azure Cognitive Services | deepseek-r1 | Custom | - | - | - | ||
| GitHub Models | deepseek/deepseek-r1 | Openai chat | - | - | - | - | - |
| iFlow | deepseek-r1 | Openai chat | - | - | - | ||
| Kilo Gateway | deepseek/deepseek-r1 | Openai chat | - | - | - | ||
| Merge Gateway | deepseek/deepseek-r1 | Openai chat | - | - | - | ||
| NanoGPT | deepseek-r1 | Openai chat | - | - | |||
| OpenRouter | deepseek/deepseek-r1 | Openai chat | - | - | - | ||
| Qiniu | deepseek-r1 | Openai chat | - | - | - | ||
| Snowflake Cortex | deepseek-r1 | Openai chat | - | - | - | - | - |
| Vercel AI Gateway | deepseek/deepseek-r1 | Openai chat | - | - | - |
ModelRegistry · 目录更新时间 2026-11-05