Apache TVM

license

Apache-2.0(OSI)

source

public

core features withheld

no

The LICENSE file contains the Apache-2.0 text and the README states the same license. The repository is public and the README describes no paid tier, enterprise edition or license-gated build, so the core is ungated.

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

  Top-level CMakeLists.txt builds each backend from the repository through plain build options: "Build with CUDA", "Build with Metal", "Build with ROCM" and "Build with LLVM, can be set to specific llvm-config path" all default OFF. Searched the body for enterprise, license key, paid, commercial and proprietary: none.

• https://raw.githubusercontent.com/apache/tvm/main/LICENSE recorded 2026-09-30

  LICENSE file is the verbatim Apache-2.0 text, opening with "Apache License" and "Version 2.0, January 2004" and "TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION".

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

  README states "TVM is licensed under the [Apache-2.0](LICENSE) license." and "We aim to create an open-source project maintained and owned by the community." Searched the body for enterprise, license key, paid, commercial and proprietary: none.

PyPI downloads of `apache-tvm` are the adoption signal here, though most installs build the project from source rather than install the wheel, so this figure is a floor on real use rather than a full measure of it. No stronger usage signal is published for the project.

• https://pypistats.org/api/packages/apache-tvm/recent recorded 2026-08-18

  last_month downloads = 13,543 for apache-tvm

Apache TVM's own optimizer and code generator retarget models from multiple framework front ends to LLVM, CUDA, ROCm, Metal, OpenCL, SPIR-V and WebGPU backends. That is a full compilation path from framework model to deployable module, which the kernel libraries and quantization toolkits in this category do not cover.

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

  Top-level CMakeLists.txt compiles backend sources in the repository, including "src/backend/webgpu/codegen/*.cc" and "src/backend/metal/codegen/*.cc", with options "Build with CUDA", "Build with Metal", "Build with ROCM" and "Build with Vulkan".

• https://raw.githubusercontent.com/apache/tvm/main/docs/arch/index.rst recorded 2026-09-30

  Architecture doc says "For backends such as x86 and ARM, we use the LLVM IRBuilder to build in-memory LLVM IR." and "We can also generate source-level languages such as CUDA C and OpenCL." and lists "MetaSchedule for automated performance tuning."

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

  README describes "Apache TVM is an open machine learning compilation framework," with "Python-first development that enables quick customization of machine learning compiler pipelines." and "Universal deployment to bring models into minimum deployable modules."

• https://tvm.apache.org/docs/get_started/overview.html recorded 2026-09-30

  Overview page says "TVM supports importing models from various frameworks, such as PyTorch and ONNX for generic ML models." and that the runtime "works on mobile, edge devices or even bare metal devices."