swyx· @swyx · X·· 2 小时前AI 评分60
AI 导读
swyx 转引 DeepSeek 知乎文章称，DeepInfer 推理框架已开源，支撑 DeepSeek V4.1 Flash 大规模服务，单集群吞吐 155 万亿 tokens/天，整体收入 $38 per card per hour（按错峰价格）。作者按 GPU 时价约 $5-10 估算，认为 DeepSeek 在其定价下有约 80% 的服务毛利。
正文 · 原文
deepseek posted way more alpha on their zhihu page, if you only read twitter you are reading the laowai approved stuff
"Based on this framework, we have implemented a high-throughput, low-latency large-scale inference service for DeepSeek V4.1 Flash. Currently, the service throughput of a single cluster is 155 trillion tokens/day, and the overall revenue is $38 per card per hour (based on off-peak prices)."
it costs roughtly $5-10 per gpu hour so depseek is claiming 80% serving margin at THEIR prices lololololol
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 http://github.com/deepseek-ai/DeepInfer 1/4在 X 查看被引用的帖子
来源：swyx · x.com