modular/modular
A repo that open-sources both a new programming language and the engine to run AI models faster
modular/modular bundles Mojo, a new programming language, with MAX, a framework for running and serving AI models, all in one open-source repository. It includes the Mojo compiler, Mojo standard library, MAX's hardware-accelerated operator library, an OpenAI-compatible inference server, and Python-based model pipelines. The project has 27441 stars and is written mainly in Mojo.
What it does
- Modular open-sourced the Mojo language and the MAX framework together, aiming for a unified platform for AI development and deployment.
- The repo is organized into the Mojo compiler (/KGEN), the Mojo standard library, MAX's accelerator library (/max/kernels), an OpenAI-compatible inference server, Python-based model pipelines, and code examples.
- External contributions are welcomed for the Mojo standard library, MAX accelerator library, MAX model architectures, examples, and docs, but not yet for the Mojo compiler itself.
- The repo's code is licensed under Apache License v2.0 with LLVM exceptions, while using and distributing MAX falls under a separate Modular Community License.
- The project runs community channels including Discord, a forum, meetup groups, community calls, and YouTube.
Why it matters
It gives developers a single ecosystem to write AI code and run it efficiently, instead of stitching together separate languages and inference engines. Because the accelerated kernels and an OpenAI-compatible server are open source, teams building their own inference infrastructure or model pipelines have real code to build on.
Terms in this repo
- Mojo · A new programming language created by Modular, the main language of this repo
- MAX · Modular's framework for running and serving AI models
- compiler · A program that translates human-written code into a form computers can execute, here for Mojo
- OpenAI-compatible endpoint · A server interface that accepts requests the same way OpenAI's API does, so existing OpenAI client code can be reused
- Apache License v2.0 · An open-source license that permits free use and modification of the code
Repository description (English)
The Modular Platform (includes MAX & Mojo)
Open on GitHubTrending repos
- cathrynlavery/diagram-designA skill that makes AI coding tools draw magazine-quality diagrams instead of generic rounded boxes
- public-apis/public-apisA giant crowd-curated directory of free APIs for developers
- semantica-agi/semanticaAn open-source graph infrastructure that lets AI agents show their work, not just their answers
- cactus-compute/needleA 14MB AI model small enough to run tool-calling on a phone or watch, without internet
- unslothai/unslothA desktop app that lets you run and train AI models on your own computer, no coding required
- macro-inc/macroAn all-in-one workspace where email, chat, docs, tasks, and CRM are cross-linked and share one AI memory
- harry0703/MoneyPrinterTurboAn open-source tool that turns a single topic or keyword into a finished short video, complete with script, footage, subtitles, and music
- basecamp/omarchyA ready-made, opinionated Linux setup built by DHH
Latest from METAL LAB
- 그록, 암호문에 숨긴 명령만으로 대화·위치정보 유출
- Cursor lets agents handle long-running tasks with new "/goal" command
- NVIDIA releases Switchyard, an LLM routing proxy for coding agents
- Caveman Cuts Claude Code Token Usage by 33% Using Caveman-Speak
- Apple Keeps 2027 Target for Camera AirPods, Qwen3.8-27B Makes 80 Tool Calls Locally