DeepSeek-V3
DeepSeek released the 文本 model · Launched on 2024-12
Open Source
文本
开源
API 可用
模型简介
DeepSeek-V3 is DeepSeek's flagship MoE model with a total of 671B parameters, activating only 37B parameters per inference. It uses Multi-head Latent Attention (MLA) and DeepSeekMoE architecture, with a training cost of only $5.57 million, yet achieving performance comparable to GPT-4o and Claude 3.5 Sonnet. It excels particularly in mathematics and coding tasks, and is one of the most influential models in the open-source community.
核心特性
- MoE 高效架构
- MLA 注意力机制
- 极低训练成本
- 完全开源
核心优势
- 性价比极高
- 数学和编码能力突出
- 开源可商用
技术规格
参数量
671B (37B 激活)
上下文窗口
128K tokens
API 支持
Yes
开源状态
Open Source
发布日期
2024-12
提供商
DeepSeek
价格信息
输入 ¥1.00/1M tokens，输出 ¥4.00/1M tokens（API）；开源免费
基准测试
MMLU 88.5%, HumanEval 93.5%, MATH 90.2%, 性能接近 GPT-4o