Overview of Qualcomm® AI Hub Workbench

• Getting started
  • Installation
  • Quick example (PyTorch)
• How it works
  • What does Qualcomm® AI Hub Workbench measure?

Examples

• Compiling Models
  • Compiling PyTorch to TensorFlow Lite
  • Compiling PyTorch model to a QNN Model Library
  • Compiling PyTorch model to a QNN DLC
  • Compiling PyTorch model to a QNN Context Binary
  • Compiling to a Precompiled QNN ONNX
  • Compiling PyTorch model for the ONNX Runtime
  • Compiling ONNX models to TensorFlow Lite or QNN
  • Compiling models quantized with AIMET to TensorFlow Lite or QNN
• Profiling Models
  • Profiling a previously compiled model
  • Profiling a PyTorch model
  • Profiling a TorchScript model
  • Profiling an ONNX model
  • Profiling an ONNX model on a device that does not support floating point
  • Profiling a QNN DLC
  • Profiling a QNN Context Binary
  • Profiling a TensorFlow Lite model
  • Profiling a model on multiple devices
  • Uploading a model for profiling
  • Profiling a previously uploaded model
• Running Inference
  • Running Inference with a TensorFlow Lite model
  • Running Inference with QNN Model Libraries, DLCs, and Context Binaries
  • Verify model accuracy on-device using an inference job
  • Running inference using a previously uploaded dataset and model
• Quantization
  • Overview
  • Preparing the source model
  • Quantizing an ONNX model
  • Compiling a quantized model
  • Quantization options
  • Benchmarking quantized model performance
• Linking
  • Variable input size example
• Devices
  • Device Selection
  • Device Filtering
• Working with Jobs
  • Querying Jobs
  • Profile Jobs
  • Sharing Jobs
• Working with Frameworks
  • Querying Frameworks
  • Submitting Jobs with Framework Versions
  • Version Selection
• Command Line Interface
  • Examples
• Deployment
  • Create Deployable ONNX models
  • Qualcomm® AI Hub Apps

API Documentation

• API documentation
  • Core API
  • Managed objects
  • Exceptions
  • Common Options
  • Compile Options
  • Quantize Options
  • Link Options
  • Profile & Inference Options
  • ONNX Runtime Options
  • ONNX Runtime QNN Execution Provider options
  • ONNX Runtime DirectML Execution Provider options
  • TensorFlow Lite Options
  • TensorFlow Lite Delegate Options for Qualcomm® AI Engine Direct
  • TensorFlow Lite Delegate Options for GPUv2
  • TensorFlow Lite Delegate Options for NNAPI
  • Qualcomm® AI Engine Direct Options

FAQ

• Frequently Asked Questions
  • General
  • Model Formats
  • Passing Options
  • Common job failures and known issues
  • Deployment

Release Notes

• Release Notes
  • Released Nov 4, 2025
  • Released Oct 27, 2025
  • Released Oct 22, 2025
  • Released Oct 6, 2025
  • Released Sept 29, 2025
  • Released Sept 22, 2025
  • Released Sept 8, 2025
  • Released Sept 2, 2025
  • Released Aug 22, 2025
  • Released Aug 11, 2025
  • Released July 28, 2025
  • Released July 14, 2025
  • Released June 30, 2025
  • Released June 16, 2025
  • Released June 2, 2025
  • Released May 14, 2025
  • Released May 5, 2025
  • Released April 22, 2025
  • Released April 8, 2025
  • Released March 25, 2025
  • Released March 10, 2025
  • Released February 24, 2025
  • Released Febuary 10, 2025
  • Released January 22, 2025
  • Released January 6, 2025
  • Released December 13, 2024
  • Released November 25, 2024
  • Released November 11, 2024
  • Released October 28, 2024
  • Released October 14, 2024
  • Released October 7, 2024
  • Released September 23, 2024
  • Released September 11, 2024
  • Released August 26, 2024
  • Released August 12, 2024
  • Released July 29, 2024
  • Released July 15, 2024
  • Released July 1, 2024
  • Released June 17, 2024
  • Released June 4, 2024
  • Released May 17, 2024
  • Released May 6, 2024
  • Released April 22, 2024
  • Released April 8, 2024
  • Released March 25, 2024
  • Released March 11, 2024
  • Released February 28, 2024