DeepSeek 以 Apache 2.0 开源推理引擎 DeepInfer
DeepSeek just open-sourced DeepInfer, the inference engine behind its production services, under the Apache 2.0 license.
DeepSeek 以 Apache 2.0 许可证开源了其生产服务所用的推理引擎 DeepInfer，8 小时内获得 6.3K+ stars。该版本只支持 DeepSeek-V4.1-Flash，运行在 NVIDIA SM100 GPU 或华为昇腾 950 芯片上，覆盖离线推理而非完整服务栈，缺少 KV cache、prefill 与 decode 分离以及副本间负载均衡。随附的加速层包含 DSpark，可提前草拟最多 5 个 token 并用置信度估计安排验证；bounded replay 只重算 140 token 后缀以覆盖滑动注意力窗口，从而压缩 prefill。配套库 deepseek-recipe 将把 Chat Completions、Responses 和 Messages 请求转换为与官方 API 行为一致的 DeepSeek V4 与 V4.1 提示词。
推荐理由
AI 评分 62
原文说明了 DeepInfer 只覆盖离线推理与加速层，并列出缺失的 KV cache、PD 分离和负载均衡，读者可据此判断它与完整服务栈的差距。
主题
正文
当前显示原文，中文全文译文尚未就绪。
DeepSeek just open-sourced DeepInfer, the inference engine behind its production services, under the Apache 2.0 license.
and it got 6.3K+ stars in 8 hours
The release runs only DeepSeek-V4.1-Flash, on NVIDIA SM100 GPUs or Huawei Ascend 950 chips, and covers offline inference rather than a complete serving stack.
It omits the parts that turn an engine into a service: a KV cache system, prefill and decode disaggregation, and load balancing across replicas.
What ships is the speed layer, including DSpark, which drafts up to 5 tokens ahead and uses confidence estimates to schedule their verification.
Bounded replay trims prefill by recomputing only a 140-token suffix sized to cover the sliding attention window.
A companion library, deepseek-recipe, will turn Chat Completions, Responses and Messages requests into DeepSeek V4 and V4.1 prompts that match the official API’s behaviour.