Footer

Your go-to destination for discovering trending open-source projects and uncovering the insights that matter

Legal

  • Terms and conditions
  • Privacy policy

More

  • Pricing
  • Advertise
  • onedump
  • gosnakego
  • Contact us

Social

  • X (Twitter)
  • GitHub
Trendshift owl
What they build speaks louder than words
© 2026 Trendshift, created by Julian Li with ❤️
SignalAdvertise
Trendshift
SignalAdvertise
Sign in
Trendshift owl

Discover trends that matter

DailyWeeklyMonthlyYearly
Live mentionsTopics
RepositoriesDevelopers
InsightsStats
Busbar
Busbar
AxiomCode
AxiomCode
Bifrost
Bifrost

deepseek-ai/DeepInfer

DeepInfer: High-performance, large-scale inference framework for DeepSeek models

Visit GitHub
Python
7.5k
104
1 contributors
Apache License 2.0
FeaturedAdvertise to 180K+ monthly visitors
Busbar

Control what AI can do before it acts

AxiomCode

Give Al Agents a Deeper Understanding of Your Codebase with a Code Graph

Bifrost

Enterprise AI Gateway Built For Scale: Route, Govern, and Secure All Your AI Traffic

Momentum

How fast deepseek-ai/DeepInfer is gaining stars, and where it lands on Trendshift and GitHub Trending

Activity and trending

Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending

Oct 19 – Oct 20, 2026

Stars +5,192 in 2 days

02K4K6K
5,192

Forks +0 in 2 days

01

Merged PRs 0 in 2 days

01

Trendshift Trendingwhere the repository ranked each day it made the top 25

Not in the Trendshift top 25 in this period

Oct 19Oct 20
The left and right arrow keys step from day to day; Home and End go to the first and the last

Stars, forks, merged pull requests and Trendshift Trending rank, by day
DayStars gainedForks gainedMerged PRs
Mon, Oct 19, 2026+0+00
Tue, Oct 20, 2026+5,192+00
Total+5,192+00
  • 1#1 on the list
  • 7Rank that day
  • Not trending

Social mentions

Recent discussions about this repository across the web
  1. Oct 205 posts10 posts
    • DeepSeek 把推理引擎和编码组件开源了🎉🎉 DeepInfer:让模型跑更快,支持推测解码、并行计算、结构化输出,面向 NVIDIA SM100 与昇腾 950 📷 deepseek-recipe:把接口转换为对话格式,让社区部署与官方 API 保持一致 📷 前者管速度,后者管一致性

      Read full on X, posted by @ytiomin

      @ytiomin · x.com

    • DeepSeek 开源了 DeepInfer,他们线上服务用的推理框架,Apache 2.0,目前支持 V4.1-Flash,跑在 NVIDIA SM100 和华为 Ascend 950 上 不过这次放出的是离线推理部分,README 写明 KV cache 系统、PD 分离、副本间负载均衡都不在里面

      Read full on X, posted by @timemate1

      @timemate1 · x.com

    • DeepSeek开源了自家推理框架DeepInfer。 专为V4.1-Flash设计,高吞吐低延迟。 目的是降低算力成本,让AI更便宜、更绿色。 生产级代码直接公开,开发者必看: #AI #OpenSource

      Read full on X, posted by @freesyscxy

      @freesyscxy · x.com

    • 前脚美国国土安全部刚把 DeepSeek 踢出联邦机构,后脚 DeepSeek 又开源了一个大件:这回不是模型,是自家线上跑 V4.1-Flash 的推理框架 DeepInfer。 - 时间:10 月 20 日傍晚(北京时间 17:30 左右)官方发推,Apache 2.0 协议,可商用,GitHub 几个小时就 7700+ star - 定位:DeepSeek…

      Read full on X, posted by @sanyanzhumo

      @sanyanzhumo · x.com

    • DeepSeek 又悄悄放了个大招。 deepseek-ai/DeepInfer:高性能大模型推理框架(仓库简介原话)。★7195,日均涨星 +1706——4 天攒出这个体量,涨速全榜第二。出品方是 DeepSeek 官方组织账号,推理侧工程能力正在往外摊。

      Read full on X, posted by @grant5407

      @grant5407 · x.com

    • 同时支持英伟达和华为昇腾:@deepseek_ai 开源推理框架 DeepInfer 1️⃣ 官方称其支撑 V4.1-Flash 大规模线上服务,开源为帮社区降低成本和能耗 2️⃣ 支持英伟达 SM100 GPU、华为昇腾 950 NPU,Apache 2.0 协议 3️⃣ 本次为离线推理版,KV 缓存等需自建;同步开源 deepseek-recipe #DeepSeek #AI

      Read full on X, posted by @GRS815

      @GRS815 · x.com

    • 🚨 卧槽!大模型又要开始新一轮降价潮! 另外那些天天喊着 DeepSeek 全靠蒸馏 的人,可以闭嘴去读源码了! 梁子今天直接把低价的看家秘密武器——DeepInfer 推理框架正式开源! 🔥硬核底层技术: 1. 集成了 FlashMLA、DeepGEMM、DeepEP 等全套自研硬核算子 2. 支持 Attention DP / 平衡 CP,以及 MoE 的 EP+EPLB 负载均衡 3.…

      Read full on X, posted by @Lonely__MH

      @Lonely__MH · x.com

    • DeepSeek 开源了自己模型使用的高性能可大规模拓展的推理框架,短短 八个小时已经获得 6000+ Star

      Read full on X, posted by @jiamu_x

      @jiamu_x · x.com

    • DeepSeek 把自家推理框架 DeepInfer 开源了,还把线上账本摊开:平均 96% 的输入命中缓存,缓存留 3 到 7 天,每张卡每小时收入 38 美元。按官网价算,一个 77K 的请求,输入费从 7.7 分降到不到半分钱。难怪敢卖这么便宜,吃饭的家伙都给了 👀

      Read full on X, posted by @finalbsd

      @finalbsd · x.com

    • DeepSeek 把内部的推理框架 DeepInfer 全面开源了,同时放出了生产级 Token 编解码组件 deepseek-recipe。 以前想复刻这条链路,只能在 vLLM 上自己改,现在官方把跑 V4.1-Flash 的生产栈直接摆了出来。 地址在 2.0;recipe 是 MIT。 没有一行装完的命令,得从源码编:Linux、GCC 13+、Python 3.12、CUDA…

      Read full on X, posted by @berryxia

      @berryxia · x.com