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

SignalAdvertise
DailyWeeklyMonthlyYearly
Live mentionsTopics
RepositoriesDevelopers
InsightsStats
Bifrost
Bifrost

deepseek-ai/DeepJIT

A lightweight library for xPU kernel JIT compilation

Visit GitHub
C++
450
74
3 contributors
MIT License
AchievementsView all →
FeaturedAdvertise to 180K+ monthly visitors
Bifrost

Fastest Enterprise AI Gateway for Routing, Governing, and Securing all your AI Traffic

Momentum

How fast deepseek-ai/DeepJIT 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

Sep 9 – Nov 5, 2026

Stars +413 in 58 days

0100200
168

Forks +74 in 58 days

01020
16

Merged PRs 10 in 58 days

012
2

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

Top 25 in C++

26
Sep 9Sep 17Sep 25Oct 3Oct 11Oct 19Oct 27Nov 4
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 PRsC++
Wed, Sep 9, 2026+0+00Not trending
Thu, Sep 10, 2026+168+161#2
Fri, Sep 11, 2026+66+60#6
Sat, Sep 12, 2026+17+10Not trending
Sun, Sep 13, 2026+10+00Not trending
Mon, Sep 14, 2026+11+30Not trending
Tue, Sep 15, 2026+13+00Not trending
Wed, Sep 16, 2026+3+00Not trending
Thu, Sep 17, 2026+1+00Not trending
Fri, Sep 18, 2026+0+10Not trending
Sat, Sep 19, 2026+0+00Not trending
Sun, Sep 20, 2026+0+00Not trending
Mon, Sep 21, 2026+3+00Not trending
Tue, Sep 22, 2026+0+41Not trending
Wed, Sep 23, 2026+4+21Not trending
Thu, Sep 24, 2026+1+00Not trending
Fri, Sep 25, 2026+0+00Not trending
Sat, Sep 26, 2026+2+00Not trending
Sun, Sep 27, 2026+2+00Not trending
Mon, Sep 28, 2026+8+11Not trending
Tue, Sep 29, 2026+1+00Not trending
Wed, Sep 30, 2026+11+12Not trending
Thu, Oct 1, 2026+2+20Not trending
Fri, Oct 2, 2026+2+00Not trending
Sat, Oct 3, 2026+2+10Not trending
Sun, Oct 4, 2026+1+20Not trending
Mon, Oct 5, 2026+0+00Not trending
Tue, Oct 6, 2026+1+00Not trending
Wed, Oct 7, 2026+0+10Not trending
Thu, Oct 8, 2026+4+00Not trending
Fri, Oct 9, 2026+4+22Not trending
Sat, Oct 10, 2026+4+00Not trending
Sun, Oct 11, 2026+2+10Not trending
Mon, Oct 12, 2026+2+10Not trending
Tue, Oct 13, 2026+2+00Not trending
Wed, Oct 14, 2026+1+00Not trending
Thu, Oct 15, 2026+2+30Not trending
Fri, Oct 16, 2026+0+10Not trending
Sat, Oct 17, 2026+1+00Not trending
Sun, Oct 18, 2026+1+00Not trending
Mon, Oct 19, 2026+2+01Not trending
Tue, Oct 20, 2026+9+30Not trending
Wed, Oct 21, 2026+13+100Not trending
Thu, Oct 22, 2026+9+00Not trending
Fri, Oct 23, 2026+2+00Not trending
Sat, Oct 24, 2026+3+10Not trending
Sun, Oct 25, 2026+0+20Not trending
Mon, Oct 26, 2026+1+10Not trending
Tue, Oct 27, 2026+2+00Not trending
Wed, Oct 28, 2026+1+10Not trending
Thu, Oct 29, 2026+2+40Not trending
Fri, Oct 30, 2026+0+10Not trending
Sat, Oct 31, 2026+2+10Not trending
Sun, Nov 1, 2026+3+00Not trending
Mon, Nov 2, 2026+2+00Not trending
Tue, Nov 3, 2026+4+11Not trending
Wed, Nov 4, 2026+4+00Not trending
Thu, Nov 5, 2026+2+00Not trending
Total+413+7410
  • 1#1 on the list
  • 7Rank that day
  • Not trending

    • Mentions and achievements

    Where people are talking about deepseek-ai/DeepJIT, and every badge it has earned
    1. Sep 291 post

      • this was back in 2023 for their R1 model which is half the size of v4. like wdym 545% profit margin? and current v4 api prices are even lower than R1. god knows how much more they have optimized.

        Read full on X, posted by @sohamstwt

        @sohamstwt · x.com

    2. 16 days earlier
    3. Sep 131 post

      • 为什么内核库总要各自重写一遍 JIT 基础设施？DeepJIT 把这层抽成轻量 C++20 运行时，统一 CUDA 与昇腾的编译、加载和启动流程。 重点在缓存设计：源码、受跟踪头文件、编译器版本、有效参数和依赖签名都进入键值；共享缓存用临时目录、fsync 与原子重命名发布，能应对多进程多节点并发。惰性初始化也避免扩展加载时就探测设备，适合嵌入 C++ 或 PyTorch 扩展。

        Read full on X, posted by @vintcessun

        @vintcessun · x.com

    4. Sep 111 post

      • DeepJIT 是 DeepSeek 开源的轻量头文件式 C++20 JIT 运行时，同时支持 NVIDIA CUDA GPU 与华为昇腾 NPU，给扩展作者提供统一的内核编译、缓存、加载与启动接口。

        Read full on X, posted by @QingQ77

        @QingQ77 · x.com

    5. Sep 102 posts5 posts

      • DeepSeek 为 V4.1 Flash 及后续模型开源了三个基础设施项目： - deepseek-recipe API 与模型之间的适配层。负责把 Messages、Chat Completions、Responses 等请求统一成内部对话格式，再编码为 DeepSeek V4／V4.1 使用的 prompt，并将模型输出还原成对应的 API 响应。 - DeepSelect 为…

        Read full on X, posted by @verysmallwoods

        @verysmallwoods · x.com

      • DeepSeek 开源了一些新的代码仓库，方便更容易地部署 V4.1 Flash 以及后续的开源模型： deepseek-recipe 解决的是“应用怎样正确地和模型对话”。 应用发送的请求格式，与模型真正接收的输入格式之间，需要转换。它支持 Chat Completions、Responses、Messages 三类接口，将请求整理成统一对话，再编码为 DeepSeek V4/V4.1…

        Read full on X, posted by @Tao100086

        @Tao100086 · x.com

      • DeepJIT: Header-Only C++20 JIT Runtime for Nvidia CUDA and Huawei Ascend

        Read full on Hacker News

        Hacker News

      • DeepSeek 开源了一些新的代码仓库，方便更容易地部署 V4.1 Flash 以及后续的开源模型：

        Read full on X, posted by @tianyi

        @tianyi · x.com

      • Backup in case of delete /s

        Read full on X, posted by @Presidentlin

        @Presidentlin · x.com

    You might also like

    Repositories close to deepseek-ai/DeepJIT
    FeaturedAdvertise to 180K+ monthly visitors
    Bifrost

    Fastest Enterprise AI Gateway for Routing, Governing, and Securing all your AI Traffic

    mitsuba-renderer/drjit-core

    C++

    Dr.Jit — A Just-In-Time-Compiler for Differentiable Rendering (core library)

    12833

    NVIDIA/tensor-ir

    C++

    TensorIR is a lightweight NVIDIA-owned MLIR compiler frontend for expressing tensor computations and lowering them to NVIDIA CUDA Tile IR.

    14117

    bertaye/agentic-cuda-optimizer

    Python

    A langgraph based workflow with a C++ CUDA harness to optimize CUDA kernels

    374

    llnl/proteus

    C++

    Programmable JIT Compilation and Optimization for C/C++ using LLVM

    5611

    HazyResearch/ThunderKittens

    Cuda

    Tile primitives for speedy kernels

    3.8k345

    NVIDIA/cutile-python

    Python

    cuTile is a programming model for writing parallel kernels for NVIDIA GPUs

    2.2k169

    Dao-AILab/quack

    Python

    A Quirky Assortment of CuTe Kernels

    1.2k178

    NVIDIA/cuda-tile

    C++

    CUDA Tile IR is an MLIR-based intermediate representation and compiler infrastructure for CUDA kernel optimization, focusing on tile-based computation patterns and optimizations targeting NVIDIA tensor core units.

    1.1k98

    HazyResearch/HipKittens

    C++

    Fast and Furious AMD Kernels

    48185