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

Omnigraph
Omnigraph
It's a plan
It's a plan
OfinTech Desk
OfinTech Desk
Busbar
Busbar
Bifrost
Bifrost

lyogavin/airllm

AirLLM 2.8T LLM inference with single 4GB GPU

Visit GitHub
Jupyter Notebook
42.6k
4.3k
9 contributors
Apache License 2.0
AchievementsView all →
FeaturedAdvertise to 180K+ monthly visitors
Omnigraph

Versioned graph database on object storage. Git-style workflows and multimodal retrieval

It's a plan

Open-source Linear alternative

An issue tracker where AI agents take the work you assign them

OfinTech Desk

An open-source AI workspace for finance and accounting teams, in which every figure is traced to its source

Busbar

Control what AI can do before it acts

Bifrost

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

Momentum

How fast lyogavin/airllm 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 18 – Nov 17, 2026

Stars +8,081 in 61 days

05001K1.5K
1,044

Forks +858 in 61 days

0204060
54

Merged PRs 10 in 61 days

0510
10

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

Top 25 overall

1624

Top 25 in Jupyter Notebook

101512117121411310131794471291921111111333111146572376810414124463793724
Sep 18Sep 26Oct 4Oct 12Oct 20Oct 28Nov 5Nov 13
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 PRsAll languagesJupyter Notebook
Fri, Sep 18, 2026+52+70Not trending#10
Sat, Sep 19, 2026+27+60Not trending#15
Sun, Sep 20, 2026+43+90Not trending#12
Mon, Sep 21, 2026+48+60Not trending#11
Tue, Sep 22, 2026+49+70Not trending#7
Wed, Sep 23, 2026+40+50Not trending#12
Thu, Sep 24, 2026+33+70Not trending#14
Fri, Sep 25, 2026+146+130Not trending#1
Sat, Sep 26, 2026+91+110Not trending#1
Sun, Sep 27, 2026+98+170Not trending#3
Mon, Sep 28, 2026+57+120Not trending#10
Tue, Sep 29, 2026+48+80Not trending#13
Wed, Sep 30, 2026+38+40Not trending#17
Thu, Oct 1, 2026+34+120Not trending#9
Fri, Oct 2, 2026+4+70Not trendingNot trending
Sat, Oct 3, 2026+45+80Not trending#4
Sun, Oct 4, 2026+49+50Not trending#4
Mon, Oct 5, 2026+39+80Not trending#7
Tue, Oct 6, 2026+28+40Not trending#12
Wed, Oct 7, 2026+42+20Not trending#9
Thu, Oct 8, 2026+27+30Not trending#19
Fri, Oct 9, 2026+28+30Not trending#21
Sat, Oct 10, 2026+334+170Not trending#1
Sun, Oct 11, 2026+567+480Not trending#1
Mon, Oct 12, 2026+1,044+540#16#1
Tue, Oct 13, 2026+807+460#24#1
Wed, Oct 14, 2026+384+280Not trending#1
Thu, Oct 15, 2026+182+90Not trending#1
Fri, Oct 16, 2026+162+210Not trending#3
Sat, Oct 17, 2026+150+170Not trending#3
Sun, Oct 18, 2026+164+230Not trending#3
Mon, Oct 19, 2026+198+2710Not trending#1
Tue, Oct 20, 2026+351+330Not trending#1
Wed, Oct 21, 2026+261+280Not trending#1
Thu, Oct 22, 2026+186+230Not trending#1
Fri, Oct 23, 2026+141+200Not trending#4
Sat, Oct 24, 2026+86+80Not trending#6
Sun, Oct 25, 2026+106+70Not trending#5
Mon, Oct 26, 2026+119+240Not trending#7
Tue, Oct 27, 2026+208+200Not trending#2
Wed, Oct 28, 2026+125+220Not trending#3
Thu, Oct 29, 2026+121+120Not trending#7
Fri, Oct 30, 2026+103+160Not trending#6
Sat, Oct 31, 2026+82+100Not trending#8
Sun, Nov 1, 2026+77+150Not trending#10
Mon, Nov 2, 2026+117+160Not trending#4
Tue, Nov 3, 2026+60+100Not trending#14
Wed, Nov 4, 2026+47+60Not trending#12
Thu, Nov 5, 2026+78+140Not trending#4
Fri, Nov 6, 2026+59+70Not trending#4
Sat, Nov 7, 2026+57+120Not trending#6
Sun, Nov 8, 2026+89+110Not trending#3
Mon, Nov 9, 2026+53+60Not trending#7
Tue, Nov 10, 2026+62+90Not trending#9
Wed, Nov 11, 2026+73+80Not trending#3
Thu, Nov 12, 2026+58+110Not trending#7
Fri, Nov 13, 2026+68+90Not trending#2
Sat, Nov 14, 2026+66+130Not trending#4
Sun, Nov 15, 2026+72+160Not trendingNot trending
Mon, Nov 16, 2026+51+180Not trendingNot trending
Tue, Nov 17, 2026+47+00Not trendingNot trending
Total+8,081+85810
  • 1#1 on the list
  • 7Rank that day
  • Not trending
1
  • GitHub trending history

    Where the repository ranked on GitHub Trending, one dot for each day it was on a list

    All languages

    Jan 23 – Oct 13, 2026#1 on 1 day

    #1#5#10#15#20#25
    Top 10
    Feb 2026AprJunAugOct
    The left and right arrow keys step from day to day; Home and End go to the first and the last

    Mentions and achievements

    Where people are talking about lyogavin/airllm, and every badge it has earned
    1. Nov 41 post
      • Big AI on cheap hardware, free? BitNet shrinks weights to -1/0/+1 to run fast on a CPU. AirLLM fits a 70B model in ~4GB GPU, but it's slow. Which would you try? #AI

        Read full on X, posted by @AndyBuildsAI

        @AndyBuildsAI · x.com

    2. Nov 21 post
      • ⚡ نموذج حجمه 70B على بطاقة رسومية بذاكرة 4GB هذا ما يقوله مشروع AirLLM المفتوح المصدر. الفكرة بسيطة: لا تحمّل النموذج كله. تضع طبقة واحدة فقط في البطاقة، ثم تنتقل إلى التي بعدها. فلا يهم حجم النموذج…

        Read full on X, posted by @Sn0wbrave

        @Sn0wbrave · x.com

    3. Oct 311 post
      • 本周 GitHub 上涨星最猛的 10 个 AI 开源项目 项目地址放在评论区了👇

        Read full on X, posted by @cj3214567667

        @cj3214567667 · x.com

    4. 10 days earlier
    5. Oct 211 post
      • 2.8T 的 Kimi K3,airllm 说 3.72GB 显存能跑。671B 的 DeepSeek-V3 约 12GB,也就一张 RTX 4070。 办法是权重一层层从硬盘读进显卡。代价在读盘:README 没给速度,光拆 Qwen3.8-Flash-Next 就要约 360GB 硬盘。跑通可以,当聊天服务别想。

        Read full on X, posted by @AIRadarDesk

        @AIRadarDesk · x.com

    6. Oct 171 post2 posts
      • 本周 GitHub 涨星最猛的 10 个 AI 开源项目,会用 AI 的人又多了 10 件趁手的家伙👇 01 REA|逆向一切 26.9 万星,4 月才建库,单周暴涨 21.4 万。看到别家 App 某个功能做得好?跟 AI 说一句「搞清楚它的搜索怎么实现的,给我证据,再在我项目里做一个」,它会调用 Ghidra、IDA 从网页一路拆到二进制,每个结论都附证据。官方案例里,AI…

        Read full on X, posted by @ArthurLiu0725

        @ArthurLiu0725 · x.com

      • 这周 GitHub 涨得最猛的十个 AI 开源项目,会用 AI 的人手里又多了哪些家伙。 01 REA 第一个,REA,全称 Reverse Engineer Anything,逆向一切。 现在二十六万九千颗星。它四月份才建库,光这一周就涨了二十一万四千颗,今天一天又是三万多。 它火在哪?你看到别的 App 里有个功能做得特别好,以前想知道它是怎么实现的,得懂汇编、会用反编译工具,一行一行去啃。…

        Read full on X, posted by @slashui

        @slashui · x.com

    7. Oct 151 post
      • 本周 GitHub 上迅速崛起的 AI 仓库 10 选 1. rea — 把应用内部的解析工作交给 AI 代理。从应用的运行行为(动作)到程序深层部分,都能帮你追溯并解读其机制。 2. skills — 一位现场工程师在日常开发中实际使用的技能集合,来自其个人工作环境。导入到 AI 后,它就能按相同流程运行。 3. diagram-design — 为 AI…

        Read full on X, posted by @superfang119

        @superfang119 · x.com

    8. Oct 142 posts
      • 1/ VoiceStudio (63k ⭐) Fully local ElevenLabs alternative: voice cloning, voice design, video dubbing, transcription. Runs on your GPU. →

        Read full on X, posted by @maga_musk

        @maga_musk · x.com

      • 本地跑2.8T模型,显存居然可以压到4GB以下。 AirLLM说不靠量化剪枝,把Kimi K3这种量级塞进单卡4GB,DeepSeek-V3大约12GB。思路是分层加载,用时再搬,不是硬塞整模。 显卡不够用的人,瓶颈多半在加载策略,不在又买一张卡。

        Read full on X, posted by @myldao

        @myldao · x.com

    9. Oct 131 post
      • ¿Un LLM gigante en una GPU de solo 4 GB? Este repo lo hace posible. AirLLM (38241 estrellas) → Carga el modelo capa por capa, sin cuantizar → Funciona con GPUs modestas y en Mac → Compresión opcional…

        Read full on X, posted by @TellusCoop

        @TellusCoop · x.com

    You might also like

    Repositories close to lyogavin/airllm
    FeaturedAdvertise to 180K+ monthly visitors
    It's a plan

    Open-source Linear alternative

    An issue tracker where AI agents take the work you assign them

    Omnigraph

    Versioned graph database on object storage. Git-style workflows and multimodal retrieval

    Busbar

    Control what AI can do before it acts

    Bifrost

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

    OfinTech Desk

    An open-source AI workspace for finance and accounting teams, in which every figure is traced to its source

    0xSojalSec/airllm

    Jupyter Notebook

    Runs 405B LLMs on 8GB VRAM

    3.1k249

    hyperionaidev/smolmaxxing

    C

    maxxing the ability to run different LLMs on the smallest devices possible

    95040

    jitllm/jitllm

    Go

    Smart inference engine that adapts to your hardware to run models faster, even when they don’t fit in memory

    586

    Tiiny-AI/PowerInfer

    C++

    High-speed Large Language Model Serving for Local Deployment

    9.9k666

    coderredlab/runNburn

    Rust

    Memory-aware GGUF inference runtime with OpenAI-compatible serving

    282

    lvyufeng/PocketLLM

    Python

    Inference engine for large language models on consumer GPUs, with deep optimization for older hardware

    5510

    sqliteai/warp

    C

    Run the full 2.78-trillion-parameter Kimi K3 model, DeepSeek V4.1 Flash or GLM-5.3-Flash beyond available RAM by streaming activated weights directly from NVMe. A dependency-free, embeddable C inference engine.

    2.6k210

    naklecha/simple-llm

    Python

    ~1300 line, minimal, extensible LLM inference engine built from scratch.

    49338

    wici-ai/inferssd

    C++

    Run models larger than your RAM or VRAM by streaming MoE experts from SSD. Built on llama.cpp.

    1.2k107