Deepseek 开源其最快最便宜模型的推理引擎
原标题：Deepseek open-sources the inference engine that powers its fastest and cheapest AI model
AI 摘要
Deepseek 以 Apache License 2.0 协议开源了其内部推理框架 DeepInfer，该框架基于 vLLM 构建，包含优化的计算内核、多芯片分布式处理和更快的文本生成技术，用于支撑其 V4.1-Flash 模型的高效推理服务。Deepseek 同时发布了 deepseek-recipe，用于复现官方 API 的 token 处理流程，使自托管部署行为一致。公司称此举旨在帮助社区降低算力成本与能耗，但报道指出这也会对 OpenAI、Anthropic 等西方厂商形成价格压力。
正文节选
Deepseek open-sources the inference engine that powers its fastest and cheapest AI model Deepseek has open-sourced DeepInfer, its internal inference framework, under the Apache License 2.0. Inference is how a trained AI model handles requests and generates responses. DeepInfer is the serving software that lets Deepseek's V4.1-Flash model produce answers quickly at scale. The framework builds on vLLM with optimized compute kernels, distributed processing across multiple chips, and faster text gen