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

Websitebench
Websitebench
Omnigraph
Omnigraph
Busbar
Busbar
Bifrost
Bifrost

karpathy/nanochat

The best ChatGPT that $100 can buy.

#LLM training
Visit GitHub
Python
60.6k
8.8k
51 contributors
MIT License
AchievementsView all →
FeaturedAdvertise to 180K+ monthly visitors
Websitebench

Can AI agents reverse-engineer real websites?

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

Momentum

How fast karpathy/nanochat 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 17 – Nov 16, 2026

Stars +2,529 in 61 days

0200400
387

Forks +500 in 61 days

0204060
55

Merged PRs 0 in 61 days

01

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

Top 25 in Python

151019
Sep 17Sep 25Oct 3Oct 11Oct 19Oct 27Nov 4Nov 12
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 PRsPython
Thu, Sep 17, 2026+31+70Not trending
Fri, Sep 18, 2026+17+110Not trending
Sat, Sep 19, 2026+15+20Not trending
Sun, Sep 20, 2026+40+100Not trending
Mon, Sep 21, 2026+28+60Not trending
Tue, Sep 22, 2026+14+60Not trending
Wed, Sep 23, 2026+22+40Not trending
Thu, Sep 24, 2026+20+20Not trending
Fri, Sep 25, 2026+5+60Not trending
Sat, Sep 26, 2026+13+60Not trending
Sun, Sep 27, 2026+12+20Not trending
Mon, Sep 28, 2026+18+40Not trending
Tue, Sep 29, 2026+23+70Not trending
Wed, Sep 30, 2026+23+30Not trending
Thu, Oct 1, 2026+24+10Not trending
Fri, Oct 2, 2026+14+50Not trending
Sat, Oct 3, 2026+12+70Not trending
Sun, Oct 4, 2026+20+40Not trending
Mon, Oct 5, 2026+7+60Not trending
Tue, Oct 6, 2026+14+30Not trending
Wed, Oct 7, 2026+22+00Not trending
Thu, Oct 8, 2026+26+00Not trending
Fri, Oct 9, 2026+28+50Not trending
Sat, Oct 10, 2026+25+70Not trending
Sun, Oct 11, 2026+16+50Not trending
Mon, Oct 12, 2026+29+50Not trending
Tue, Oct 13, 2026+21+80Not trending
Wed, Oct 14, 2026+30+80Not trending
Thu, Oct 15, 2026+19+60Not trending
Fri, Oct 16, 2026+37+30Not trending
Sat, Oct 17, 2026+22+50Not trending
Sun, Oct 18, 2026+31+40Not trending
Mon, Oct 19, 2026+22+60Not trending
Tue, Oct 20, 2026+41+70Not trending
Wed, Oct 21, 2026+19+120Not trending
Thu, Oct 22, 2026+19+40Not trending
Fri, Oct 23, 2026+26+110Not trending
Sat, Oct 24, 2026+25+30Not trending
Sun, Oct 25, 2026+21+80Not trending
Mon, Oct 26, 2026+41+110Not trending
Tue, Oct 27, 2026+76+160Not trending
Wed, Oct 28, 2026+40+60Not trending
Thu, Oct 29, 2026+16+130Not trending
Fri, Oct 30, 2026+30+100Not trending
Sat, Oct 31, 2026+49+90Not trending
Sun, Nov 1, 2026+298+550#15
Mon, Nov 2, 2026+387+520#10
Tue, Nov 3, 2026+334+320#19
Wed, Nov 4, 2026+64+100Not trending
Thu, Nov 5, 2026+51+90Not trending
Fri, Nov 6, 2026+27+100Not trending
Sat, Nov 7, 2026+24+70Not trending
Sun, Nov 8, 2026+14+30Not trending
Mon, Nov 9, 2026+28+60Not trending
Tue, Nov 10, 2026+27+50Not trending
Wed, Nov 11, 2026+35+70Not trending
Thu, Nov 12, 2026+58+120Not trending
Fri, Nov 13, 2026+26+60Not trending
Sat, Nov 14, 2026+9+70Not trending
Sun, Nov 15, 2026+23+50Not trending
Mon, Nov 16, 2026+21+00Not trending
Total+2,529+5000
  • 1#1 on the list
  • 7Rank that day
  • Not trending
2
  • GitHub trending history

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

    Feb 2 – Nov 3, 2026Best #4

    #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 karpathy/nanochat, and every badge it has earned
    1. Nov 151 post
      • nanochat packs pretrain, SFT and RL into one readable repo. Karpathy's speedrun script reportedly trains a small chat model on a single 8-GPU box in hours, so you can read every stage, change one and…

        Read full on X, posted by @version6media

        @version6media · x.com

    2. Nov 133 posts
      • 整理了一条从地基到 Agent,再到 RL 和论文的学习路线。前面是课和书,中间是能自己跑的项目,最后是读论文、写论文的方法。 1. 先把地基打好 Karpathy 的 Neural Networks: Zero to Hero,适合从零把神经网络,以及 GPT 的最小实现走通 他现在把从零训练这条线放在了 nanochat,nanoGPT 已经停更 斯坦福 CS336 Language…

        Read full on X, posted by @Zwx050616

        @Zwx050616 · x.com

      • 整理了一条从地基到 Agent,再到 RL 和论文的学习路线。前面是课和书,中间是能自己跑的项目,最后是读论文、写论文的方法。 1. 先把地基打好 Karpathy 的 Neural Networks: Zero to Hero,适合从零把神经网络,以及 GPT 的最小实现走通 他现在把从零训练这条线放在了 nanochat,nanoGPT 已经停更 斯坦福 CS336 Language…

        Read full on X, posted by @Zwx050616

        @Zwx050616 · x.com

      • 炸裂!!OpenAI 的联合创始人 Karpathy 在Github 开源了6个项目,总计超30万颗星! 这位创始人把从训练 GPT、到让模型自己做研究的一整套材料,都开源在了 GitHub 上⬇️: 1. autoresearch · 近 10 万星 能自己连夜跑实验,只把有用的结果留下 2. nanoGPT · 6.5 万星 从零开始训练一个自己的 GPT,很多人入门都是看这个仓库 3.…

        Read full on X, posted by @HeLi8866

        @HeLi8866 · x.com

    3. Nov 121 post
      • 🤯卧槽,Karpathy 把自己整套 AI 课,免费开源到 GitHub 了!?! 按学习顺序,一共 7 个仓库: → nn-zero-to-hero,从零开始搭神经网络的视频课 → micrograd,用很少的代码把 AI 怎么学会的那套数学讲清楚 → makemore,一个会自己编新名字的小模型 → minbpe,讲模型怎样把句子切成 token → nanoGPT,可以自己动手训练一个…

        Read full on X, posted by @HeLi8866

        @HeLi8866 · x.com

    4. Nov 61 post
      • 2019年训一个GPT-2要花4.3万美元,Karpathy的nanochat把它压到了48美元、约2小时。分词、预训练、微调、评估、推理全在一套极简代码里,一个--depth参数调模型大小,8张H100就能从零训出能聊天的模型。想真正理解LLM是怎么训出来的,跑一遍比看十篇教程管用。

        Read full on X, posted by @GoodMan_Lee

        @GoodMan_Lee · x.com

    5. Nov 42 posts
      • 100 美元,59983 个星,今天又涨了 321。 karpathy/nanochat 给自己的介绍只有一句:The best ChatGPT that $100 can buy。卖点没放在参数量或榜单分数上,放在了价格上。对想自己做一个聊天模型的人来说,这个门槛就不用按 GPU 集群来算了,按的是一笔个人掏得起的钱。 注意那个 best。它说的是 100 美元能买到的里面最好的,没说能比得上…

        Read full on X, posted by @iluciddreaming

        @iluciddreaming · x.com

      • 今天 GitHub 被「AI操控电脑+AI游戏工作室+百元ChatGPT」联手屠榜了!3 个星标暴增最狠的开源项目,建议收藏👇 1. vercel-labs/agent-desktop Vercel 出的命令行工具,让 AI 像人一样操作电脑桌面。 亮点:点鼠标、敲键盘的活,以后可以交给 AI。 🔗 一天 +2,243 星,总星 9,555 2.…

        Read full on X, posted by @GraceMcmenamin

        @GraceMcmenamin · x.com

    6. Nov 32 posts
      • Karpathy 新搞的 nanochat 确实挺有意思。 它把大模型训练的全流程全包了。 从分词、预训练、微调到推理,一个 8 卡 H100 节点全搞定。 2019 年花四万多刀才能训出来的 GPT-2,现在花 48 块钱、两小时就能跑完。 改个复杂度参数,超参数自动就算好了。 直接丢进命令行就能开聊。

        Read full on X, posted by @musen9527

        @musen9527 · x.com

      • Karpathy 的 nanochat,一套最简的大模型训练框架:分词、预训练、SFT、强化学习到评测,全在一个仓库,主要调 --depth 一个参数。 按 README 的算法,8 卡 H100 跑约 2 小时、48 美元左右,就能训出 GPT-2 水平的模型。 想看懂 ChatGPT 怎么炼成的,可以读读它的代码。 👉

        Read full on X, posted by @andrea_denkse

        @andrea_denkse · x.com

    You might also like

    Repositories close to karpathy/nanochat
    FeaturedAdvertise to 180K+ monthly visitors
    Websitebench

    Can AI agents reverse-engineer real websites?

    Busbar

    Control what AI can do before it acts

    Omnigraph

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

    Bifrost

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

    karpathy/autoresearch

    Python

    AI agents running research on single-GPU nanochat training automatically

    100.1k13.8k

    karpathy/nanoGPT

    Python

    The simplest, fastest repository for training/finetuning medium-sized GPTs.

    64.8k11.2k

    lm-sys/FastChat

    Python

    An open platform for training, serving, and evaluating large language models. Release repo for Vicuna and Chatbot Arena.

    39.5k4.7k

    yangqi0/petitgpt

    Python

    A 124.6M LLM trained from scratch on 13B tokens on a single RTX 4090 — tokenizer, pretraining, post-training, evaluation, and reproducible inference.

    7318

    ystemsrx/mini-nanoGPT

    Python

    One-click training of your own GPT. Training a GPT has never been easier for beginners. / 一键预训练+SFT一个属于自己的LLM,0基础训练GPT原来可以这么简单?

    38754

    OpenPipe/OpenPipe

    TypeScript

    Turn expensive prompts into cheap fine-tuned models

    2.8k185

    imoneoi/openchat

    Python

    OpenChat: Advancing Open-source Language Models with Imperfect Data

    5.5k435

    vukrosic/llm-research-kit

    Python

    LLM that can be trained on 1 or more GPUs for research.

    5917

    naklecha/simple-llm

    Python

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

    49338