Apache TVM

aid: apache-tvm
name: Apache TVM
description: Apache TVM is an open-source compiler framework for deep learning that provides performance portability across
  diverse hardware backends including CPUs, GPUs, FPGAs, and specialized accelerators (ARM, NVIDIA, AMD, Qualcomm). It automatically
  optimizes deep learning models from frameworks like TensorFlow, PyTorch, ONNX, MXNet, and Keras for deployment on edge and
  cloud targets. TVM is an Apache Software Foundation top-level project.
type: Index
position: Consumer
access: 3rd-Party
image: https://kinlane-images.s3.amazonaws.com/shared/apis-json/apis-json-logo.jpg
tags:
- AI
- Compiler
- Deep Learning
- Edge Computing
- Model Optimization
- Open Source
created: '2026-03-16'
modified: '2026-04-19'
url: https://raw.githubusercontent.com/api-evangelist/apache-tvm/refs/heads/main/apis.yml
specificationVersion: '0.19'
apis:
- aid: apache-tvm:apache-tvm-python-api
  name: Apache TVM Python API
  description: The TVM Python API provides a comprehensive interface for model compilation, optimization, and deployment.
    Key modules include tvm.relay for defining and optimizing computational graphs, tvm.auto_scheduler for auto-tuning operator
    schedules, tvm.micro for microcontroller deployment (MicroTVM), and tvm.rpc for remote deployment and profiling. The tvmc
    command-line tool provides a simplified interface for common TVM workflows.
  humanURL: https://tvm.apache.org/docs/reference/api/python/
  tags:
  - Python
  - Deep Learning
  - Model Optimization
  - Compiler
  properties:
  - type: Documentation
    url: https://tvm.apache.org/docs/reference/api/python/
  - type: SDKs
    url: https://pypi.org/project/apache-tvm/
    title: Python Package (PyPI)
- aid: apache-tvm:apache-tvm-rpc-api
  name: Apache TVM RPC API
  description: The TVM RPC (Remote Procedure Call) system enables remote compilation, deployment, and profiling of optimized
    models on target devices. It provides server/client APIs for uploading and executing compiled modules on remote hardware,
    tracking performance metrics, and running AutoTVM/AutoScheduler tuning jobs against real hardware targets.
  humanURL: https://tvm.apache.org/docs/how_to/work_with_microtvm/
  tags:
  - RPC
  - Remote
  - Profiling
  - Hardware
  properties:
  - type: Documentation
    url: https://tvm.apache.org/docs/how_to/work_with_microtvm/
common:
- type: GitHubRepository
  url: https://github.com/apache/tvm
- type: Documentation
  url: https://tvm.apache.org/docs/
- type: Portal
  url: https://tvm.apache.org/
- type: GettingStarted
  url: https://tvm.apache.org/docs/get_started/
- type: ReleaseNotes
  url: https://github.com/apache/tvm/releases
- type: Support
  url: https://discuss.tvm.apache.org/
- type: TermsOfService
  url: https://www.apache.org/licenses/
- type: Features
  data:
  - name: Multi-Framework Support
    description: Import models from TensorFlow, PyTorch, ONNX, MXNet, Keras, and other frameworks.
  - name: Hardware-Specific Optimization
    description: Automatic operator scheduling and kernel fusion for CPUs, GPUs, and custom accelerators.
  - name: Auto-Tuning
    description: AutoTVM and AutoScheduler for automated hyperparameter optimization of compute kernels.
  - name: MicroTVM
    description: Deploy optimized models on microcontrollers and bare-metal devices without an OS.
  - name: BYOC Framework
    description: Bring Your Own Codegen framework for integrating custom hardware accelerators.
  - name: Relay IR
    description: High-level intermediate representation for end-to-end model optimization.
- type: UseCases
  data:
  - name: Edge AI Deployment
    description: Deploy optimized deep learning models on edge devices and microcontrollers.
  - name: Model Serving Optimization
    description: Optimize inference performance for cloud GPU/CPU model serving.
  - name: Cross-Platform Deployment
    description: Compile a single model for multiple hardware targets from one codebase.
  - name: Custom Accelerator Integration
    description: Integrate custom AI accelerators using TVM's BYOC framework.
- type: Integrations
  data:
  - name: ONNX
    description: Import and optimize ONNX models from any ONNX-compatible ML framework.
  - name: PyTorch
    description: TorchScript to TVM compilation for PyTorch model optimization.
  - name: TensorFlow
    description: TensorFlow and TFLite model import and optimization.
  - name: NVIDIA CUDA
    description: CUDA/cuDNN backend for NVIDIA GPU kernel generation and optimization.
  - name: ARM
    description: ARM CPU (Cortex-A, Cortex-M) and ARM Mali GPU backend support.
maintainers:
- FN: Kin Lane
  email: info@apievangelist.com