IREE

license

Apache-2.0(OSI)

source

public

core features withheld

no

The LICENSE file contains the Apache-2.0 text with LLVM exceptions. The repository is public, and neither the README nor the CMake driver options mention a paid tier, enterprise edition or license key, so the core is ungated.

• https://raw.githubusercontent.com/iree-org/iree/main/CMakeLists.txt recorded 2026-09-30

  Top-level CMakeLists.txt builds each accelerator driver from the repository through plain options: "Enables the 'cuda' runtime HAL driver", "Enables the 'hip' runtime HAL driver" and "Enables the 'vulkan' runtime HAL driver". Searched the body for enterprise, license key, paid, commercial and proprietary: none.

• https://raw.githubusercontent.com/iree-org/iree/main/LICENSE recorded 2026-09-30

  LICENSE file is the Apache-2.0 text, opening with "Apache License" and "Version 2.0, January 2004" and "TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION", followed by a "LLVM Exceptions to the Apache 2.0 License" section.

• https://raw.githubusercontent.com/iree-org/iree/main/README.md recorded 2026-09-30

  README states "IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions." and describes an "MLIR](https://mlir.llvm.org/)-based end-to-end" compiler; the repository builds from source per "instructions on building from source". Searched the body for enterprise, license key, paid, commercial, proprietary and closed: none.

PyPI downloads of `iree-base-compiler` are the adoption signal here.

• https://pypistats.org/api/packages/iree-base-compiler/recent recorded 2026-08-18

  last_month downloads = 76,001 for iree-base-compiler

IREE is an MLIR-based end-to-end compiler and runtime that lowers PyTorch, JAX, TensorFlow and ONNX models to CPU, CUDA, ROCm, Vulkan and SPIR-V targets. That is a retargetable, multi-vendor compilation path comparable to Apache TVM's.

• https://raw.githubusercontent.com/iree-org/iree/main/docs/website/docs/index.md recorded 2026-09-30

  Docs home lists importing from "JAX", "ONNX", "PyTorch" and "TensorFlow and TensorFlow Lite", and accelerator support of "Vulkan", "ROCm/HIP", "CUDA" and "Metal (for Apple silicon devices)" alongside ARM, x86 and RISC-V; generating "SPIR-V kernels and Vulkan API calls" for Vulkan GPU targets.

• https://raw.githubusercontent.com/iree-org/iree/main/docs/website/docs/index.md recorded 2026-09-30

  Docs home says compilation is "performed ahead-of-time on a _host_ machine for any" combination of targets, and lists "Bare metal" and "Android" runtime platforms with "Binary size as low as 30KB on embedded systems".

• https://raw.githubusercontent.com/iree-org/iree/main/README.md recorded 2026-09-30

  README describes IREE as an "MLIR](https://mlir.llvm.org/)-based end-to-end" compiler and runtime that "lowers Machine Learning (ML) models to a unified IR" scaling from datacenter to mobile and edge deployments, and notes "IREE joins the LF AI & Data Foundation as a sandbox-stage project".