DeepSeek 发布 DeepSeek-V3，671B MoE 参数、60 tokens/秒
🚀 Introducing DeepSeek-V3 | DeepSeek API Docs
DeepSeek 发布 DeepSeek-V3，采用 671B MoE 参数、37B 激活参数，基于 14.8T 高质量 token 训练，生成速度 60 tokens/秒，比 V2 快 3 倍，API 保持兼容并开源模型与论文。API 定价在 2 月 8 日前与 V2 相同，此后调整。官方表示后续将支持多模态等功能。
推荐理由
AI 评分 88
DeepSeek-V3 给出 671B MoE、37B 激活参数与 60 tokens/秒的官方数据，可据此判断开源模型与闭源模型的差距变化。
主题
正文
当前显示原文，中文全文译文尚未就绪。
Introducing DeepSeek-V3 2024/12/26
Biggest leap forward yet
⚡ 60 tokens/second (3x faster than V2!); 💪 Enhanced capabilities; 🛠 API compatibility intact; 🌍 Fully open-source models & papers
🎉 What’s new in V3
🧠 671B MoE parameters; 🚀 37B activated parameters; 📚 Trained on 14.8T high-quality tokens
🔗 Dive deeper here:
💰 API Pricing Update
🎉 Until Feb 8: same as V2!; 🤯 From Feb 8 onwards:
Still the best value in the market! 🔥
🌌 Open-source spirit + Longtermism to inclusive AGI; 🌟 DeepSeek’s mission is unwavering. We’re thrilled to share our progress with the community and see the gap between open and closed models narrowing; 🚀 This is just the beginning! Look forward to multimodal support and other cutting-edge features in the DeepSeek ecosystem; 💡 Together, let’s push the boundaries of innovation!