AI Potluck
Model components / Inference code

MAX

Modular

MAX is an inference engine built on the Mojo programming language that serves LLMs and other models across NVIDIA, AMD, and Apple GPUs plus CPUs from a single codebase. A graph compiler optimizes model execution for the target hardware, and it serves PyTorch and Hugging Face models through OpenAI-compatible endpoints. Modular, the company founded by LLVM and Swift creator Chris Lattner, develops and distributes it.

The repository source and MAX/Mojo usage carry different licenses; the openness score follows the usage terms rather than the repository file. Verified 2026-08-09 via GitHub and the Modular Community License page.

Openness

2 high confidence
2.0
license
Modular-Community-License(non-OSI, per-device licensing and a limited redistribution grant, governs MAX/Mojo usage+distribution)
repo-license
Apache-2.0-WITH-LLVM-exception(the source in modular/modular)
source
public(~450k LOC incl MAX kernels + inference server + Mojo stdlib)
commercial-tier
Modular enterprise/managed deployment

A large public codebase whose own LICENSE file is Apache-2.0 with LLVM Exceptions, but the GitHub repository page states that "Modular, MAX and Mojo usage and distribution are licensed under the Modular Community License", which is not OSI approved. That license splits capacity by device architecture, with no cap on x86, ARM or NVIDIA PTX devices but an eight-device cap on any other accelerator type; it grants only "a limited right to redistribute certain components of the SDK"; and while the community variant is "provided free of charge", Modular "may at any time change its provision" with only commercially reasonable notice. The source is therefore readable but not unconditionally runnable or redistributable at scale, which makes this source-available rather than open source despite the product being marketed as open.

Adoption

3 low confidence
3.0

No download, user or token figure is published for MAX serving specifically. GitHub stars are the only signal available, and the 26,777 stars on modular/modular fall in the >10K stars band, level 3 on the stars scale. Confidence stays low for two reasons: a star is not a download, and modular/modular is a single repository covering MAX and Mojo together, so an unknown share of those stars belongs to the language rather than to the serving product.

Capability

4 medium confidence
4.0

The documentation records serving 500+ Hugging Face models through an OpenAI-compatible REST API, a graph compiler with operator-level fusions, and native MAX and PyTorch execution across GPUs and CPUs; the GitHub README records separate NVIDIA and AMD GPU containers, Apple-silicon GPU support, and custom operator and kernel authoring in Mojo. No standardized MLPerf or Artificial Analysis inference-engine result has been found for it, so the score rests on that feature set rather than on a benchmark.

  • https://docs.modular.com/max/intro recorded 2026-08-09

    High-performance, portable serving: Serve 500+ AI models from Hugging Face using our OpenAI-compatible REST API with industry-leading performance across GPUs and CPUs.

  • https://github.com/modular/modular recorded 2026-08-09

    We have separate containers for NVIDIA and AMD GPU environments, and a unified container that works with both.

Verified 2026-08-09