#99th of 24 in Model serving
modular
Mojo language and MAX inference framework with an OpenAI-compatible server
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 30k
- License
- custom license
- Last commit
- Oct 2026
- Last release
- Sep 2026
- Language
- Mojo
25,488 on 25 Mar 202629,951 on 14 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Monorepo for the open-source parts of the Modular Platform: the Mojo compiler and standard library, the MAX accelerator kernel library, and MAX Python model pipelines. MAX includes an inference server that exposes an OpenAI-compatible endpoint. The repo code is Apache 2.0 with LLVM Exceptions, while MAX usage and distribution fall under the Modular Community License.
Who it is for: Engineers serving models with MAX or writing GPU kernels in Mojo
Strengths
- Serving layer exposes an OpenAI-compatible endpoint, per the README
- Open kernels, Mojo stdlib and model architectures accept outside contributions
- Includes Python-based model pipelines and code examples for MAX and Mojo
- Active: last commit 2026-10-08, release 2026-09-17
Weaknesses
- Two licenses: repo is Apache 2.0 with LLVM exceptions, MAX uses Modular Community License
- Mojo compiler is not open to contributions yet
- README gives no hardware requirements or supported model list
- Broad platform repo; serving is one part, so setup is spread across separate docs
What it needs
- Models: Hugging Face models via MAX pipelines
Also in Model serving
See all 24| Rank | Project | Score |
|---|---|---|
| 1 | LocalAISelf-hosted inference server with OpenAI-compatible API and pluggable backends | 81 out of 100 |
| 2 | llama.cppC/C++ inference engine serving GGUF models over an OpenAI-compatible API | 80 out of 100 |
| 3 | OllamaLocal model runner with CLI and REST API for open-weight LLMs | 79 out of 100 |
| #4 | vLLMInference engine for Hugging Face LLMs with PagedAttention and continuous batching | 76 out of 100 |
| #5 | colibriC inference engine that runs huge MoE models by streaming experts from disk | 71 out of 100 |