ON THIS PAGE

Overview
Upcoming Events
Get Started
Developer Days Highlights
More Resources

Overview

Build Mobile and Edge AI with ExecuTorch on Arm

Build, deploy, and optimize AI applications using ExecuTorch on Arm-based platforms. Explore hands-on Learning Paths, technical articles, runnable code, and model artifacts organized around your development goals.

Upcoming Events

Upcoming Developer Events

Connect with developers, explore on-device AI, and gain hands-on experience building with ExecuTorch and Arm-based platforms.

Arm + Meta ExecuTorch Developer Meetup

November 13, 2026 | Northeastern University, Seattle, Washington

Sponsors: Arm and Meta

Join Arm and Meta for an in-person tech talk and developer meetup. Explore on-device AI development with ExecuTorch, exchange technical insights, and connect with the developer community.

Event details coming soon

Namma AI Hackathon: Build Edge AI Applications

November 14, 2026 | NASSCOM, Bangalore, India

Sponsors: Arm and IIIT-Bangalore

Put your edge AI skills into practice at the Namma AI Hackathon. Collaborate with fellow developers to build AI applications that address real-world challenges using on-device intelligence.

Explore the Hackathon

Get Started

Start Developing with ExecuTorch

Find Learning Paths, code examples, models and technical guidance to build, deploy and optimize AI applications on Arm.

Build Computer Vision Applications

Image classification, object detection, and image segmentation.

Learn More

Build Speech and Audio Applications

Speech recognition, voice activity detection, and text-to-audio.

Learn More

Build Large Language Models and Chatbots

Android chatbots, local language models, and edge Linux.

Learn More

Deploy AI on Embedded Devices

Arm Ethos-U NPUs, Arm Cortex-M devices, and development kits.

Learn More

Profile and Optimize AI Inference

Model analysis, profiling, backend selection, and CPU optimization.

Learn More

Explore More AI Development Resources

Explore the Arm AI Portal to find models, code examples, and developer resources for a wider range of AI applications.

Explore Arm AI Portal

Develop Computer Vision Applications

Develop image classification, segmentation, and visual recognition applications with ExecuTorch on Arm-based devices.

LEARNING PATHS
BLOGS
CODE
MODELS AND MORE

Run image classification on an Alif Ensemble E8 DevKit using ExecuTorch and Ethos-U85

Deploy a MobileNetV2 image classification model on Cortex-M55 and accelerate inference with Ethos-U85.

Run MNIST on an Alif Ensemble E8 DevKit using ExecuTorch and Ethos-U85

Configure an Alif Ensemble E8 development kit and run handwritten digit classification with ExecuTorch and Ethos-U85 acceleration.

Run parallel vision inference on an Alif Ensemble E8 with Zephyr

Run concurrent vision inference workloads on separate Ethos-U55 and Ethos-U85 NPUs using Zephyr.

Run MobileSAM prompt segmentation on Arm Ethos-U85 with ExecuTorch

Quantize and deploy MobileSAM on Arm Corstone-320, then validate image segmentation outputs.

Edge AI on Arm: PyTorch and ExecuTorch rock-paper-scissors

Train and deploy a simple convolutional neural network to recognize rock-paper-scissors gestures on Arm-based edge devices.

Gesture-based touchless infotainment using Arm Ethos-U NPU

Explore how to build a gesture-controlled automotive infotainment application using an NXP i.MX 93 platform and Ethos-U65.

Convert American Sign Language to text using Arm Ethos-U NPU

Explore the model and deployment workflow behind a practical edge AI accessibility application.

Edge-based fall detection using Arm Ethos-U NPU

Examine how to implement an on-device fall detection application using computer vision and Arm Ethos-U.

Deploying PaddlePaddle models on Arm Ethos-U85

Follow a PaddlePaddle deployment workflow for running vision workloads on an Ethos-U85 virtual platform.

DeiT-Tiny image classification example

Explore example code to quantize, export, and run DeiT-Tiny image classification on an Ethos-U target.

MobileSAM prompt segmentation example

Use the ExecuTorch Arm examples to prepare MobileSAM, run segmentation inference, and verify generated masks.

MobileNetV3-Large INT8

Download an Arm-optimized ExecuTorch model artifact to evaluate image classification on Ethos-U85.

ResNet-18 INT8

Explore a quantized ResNet-18 model for image classification deployment on Arm Ethos-U.

DeiT-Tiny INT8

Test transformer-based image classification with an Arm-optimized DeiT-Tiny ExecuTorch artifact.

ResNet-50 INT8

Evaluate a quantized ResNet-50 model for image classification on Arm-based embedded systems.

Inception V3 INT8

Explore an Arm-optimized Inception V3 artifact for embedded image classification experiments.

Develop Speech and Audio Applications

Develop speech recognition, voice activity detection, and audio generation applications that run directly on Arm-based devices.

LEARNING PATHS
BLOGS
CODE
MODELS and MORE

Deploy Silero VAD on Arm Ethos-U with ExecuTorch

Export and deploy a stateful voice activity detection model, then compare its output with a host reference.

Generate audio with Stable Audio Open Small using ExecuTorch

Learn how to prepare and deploy a generative audio model for on-device inference across Android and macOS.

End-to-end INT8 Conformer on Arm: Training, quantization, and deployment on Ethos-U85

Learn how to train, quantize, and deploy a Conformer speech recognition model with ExecuTorch and Ethos-U85.

Conformer training and quantization examples

Access example code to train a Conformer speech recognition model and prepare it for quantized deployment.

Arm ML Embedded Evaluation Kit

Build and run embedded speech recognition examples using ExecuTorch on Arm Fixed Virtual Platforms.

ExecuTorch Conformer small

Download Conformer speech recognition weights and quantized artifacts to prepare an Ethos-U deployment.

Run LLMs and Build Chatbots

Build on-device chatbots and large language model applications for Android smartphones and Arm-based Linux systems.

LEARNING PATHS
CODE
MODELS and MORE

Build a customer support chatbot on Android with Llama and ExecuTorch

Learn how to run a conversational AI workload locally on Arm-based mobile devices and optimize inference with Arm KleidiAI.

Build an Android chat app with Llama and ExecuTorch

Create an Android chat interface and run a Llama model locally using ExecuTorch and XNNPACK.

Run Llama 3 on a Raspberry Pi 5 using ExecuTorch

Prepare and deploy Llama 3 on a Raspberry Pi 5 to explore local language model inference.

ExecuTorch Android examples

Explore Android application examples and integrate an on-device Llama chat interface into your development workflow.

ExecuTorch Llama examples

Use the ExecuTorch source repository and Llama example workflows to build and run language models on Arm-based devices.

Arm AI Portal model catalog

Find ExecuTorch-compatible model artifacts and filter by workload, target hardware, and execution backend.

Deploy AI on Embedded Devices

Build and deploy TinyML applications on Arm Cortex-M processors and Ethos-U NPUs using physical development boards and virtual platforms.

LEARNING PATHS
BLOGS
CODE
MODELS and MORE

Introduction to TinyML on Arm using PyTorch and ExecuTorch

Set up a TinyML development environment and deploy a model using ExecuTorch on Corstone-320.

Deploy ExecuTorch firmware on NXP FRDM i.MX 93 for Ethos-U65 acceleration

Explore how Linux, the Cortex-M33, and Ethos-U65 work together to deploy and accelerate AI inference on a heterogeneous Arm system.

Faster GenAI at the edge with Arm and ExecuTorch

Explore the tooling and deployment workflow for running PyTorch models through ExecuTorch on Arm Ethos-U85.

ExecuTorch Arm backend examples

Use reference workflows for model export, quantization, Vela compilation, and inference on Arm Fixed Virtual Platforms.

Minimal Ethos-U notebook

Follow a minimal example to export a model and execute it with ExecuTorch on an Ethos-U virtual platform.

CMSIS-ExecuTorch integration example

Integrate the ExecuTorch runtime into a CMSIS-based project targeting Arm embedded systems.

Ethos-U workshop for Alif E8

Build and run a MobileNetV2 inference application with ExecuTorch on an Alif Ensemble E8 development kit.

Arm AI Portal

Discover compatible AI models, code examples, and development resources for Arm-based embedded platforms.

Arm AI Portal code catalog

Find reusable sample code and reference implementations for developing AI applications on Arm.

Profile and Optimize AI Inference

Analyze model compatibility, inspect execution behavior, and optimize inference on Arm CPUs and Ethos-U NPUs.

LEARNING PATHS
BLOGS
CODE
MODELS and MORE

Visualize Ethos-U NPU performance with ExecuTorch on Arm FVPs

Run MobileNetV2 and inspect simulated CPU and Ethos-U activity to understand model execution.

Analyze Ethos-U models with Arm ML Inference Advisor

Evaluate model compatibility, operator support, and performance estimates before deploying to Ethos-U.

Inspect model artifacts and runtime profiles with Google Model Explorer

Inspect ExecuTorch graphs, identify NPU delegation boundaries, and correlate model artifacts with runtime profiling data.

Profile ExecuTorch models with SME2 on Arm

Measure inference latency and identify performance bottlenecks on Arm devices that support Scalable Matrix Extension 2 (SME2).

Benchmark a KleidiAI micro-kernel in ExecuTorch

Measure KleidiAI micro-kernel performance in ExecuTorch and examine how kernels execute on Arm64 platforms to identify optimization opportunities.

Scaling on-device AI across different Arm backends with ExecuTorch

Understand how ExecuTorch partitions models across Arm backends, lowers supported operators, and handles non-delegated execution.

Ethos-U pruning notebook

Explore model pruning techniques and prepare a smaller neural network for deployment on Ethos-U85.

Arm AI Portal code catalog

Find reusable sample code and reference implementations for developing AI applications on Arm.

Developer Days Highlights

See What Developers Built with ExecuTorch on Arm

San Francisco State University | October 17–18, 2026

At the Arm + ExecuTorch Edge AI Challenge, students built and demonstrated on-device AI applications using Alif Ensemble E8 development boards, with support from Arm and Meta technical experts.

Alfy

1st place

An interactive, animated purple bunny powered by four ExecuTorch models. Alfy responds to faces, voice commands, touch and drawings, entirely on the Alif E8 without cloud services.

Learn More

FingerSpell

2nd place

An offline ASL fingerspelling-to-text application that detects hand landmarks, recognizes letters and assembles editable text on the device display.

Learn More

WALDO

3rd place

A voice-controlled pan-and-tilt tracking turret that detects and follows people or objects, combining vision, keyword spotting and motor control on-device.

Learn More

CorteX ARM

4th place

An EEG-based application that classifies imagined left- and right-hand movements using a compact neural network deployed with ExecuTorch, exploring potential brain–computer interface applications.

Learn More

More Resources

Take your Project Further with Arm AI Portal

Discover models, sample code, and tools to support AI development across mobile, embedded, and edge platforms.

Discover AI Resources

Build with PyTorch Across Cloud, Mobile, and Edge

Explore learning paths, code examples, and technical resources to train, optimize, and deploy PyTorch models on Arm-based platforms.

Explore PyTorch on Arm

Join the Arm Developer Program

Build AI at scale on Arm with a personalized dashboard, early access tools, on-demand labs, courses and certifications, badging, and direct access to Arm experts.

Explore the Program