---
markdown_generated_at: "2026-11-10"
title: "Windows on Snapdragon"
url: "https://www.qualcomm.com/developer/windows-on-snapdragon/windows-on-snapdragon-ai"
---

# Windows on Snapdragon

![](https://s7d1.scene7.com/is/image/dmqualcommprod/Hero-Desktop?fmt=webp-alpha&qlt=85)

Build AI – powered Windows apps with Snapdragon X Series.

[Get Started](https://docs.qualcomm.com/nav/home/ai-app-development.html?product=1601111740057789)

Run concurrent on-device AI apps and agentic workloads efficiently with the Snapdragon X Series industry-leading Qualcomm® Hexagon™ NPU for faster responsiveness and reduced latency.¹

## Why developers are building AI apps on Snapdragon

The Hexagon NPU in the Snapdragon X Series delivers performance and high efficiency with up to 80 TOPS of AI processing via its 64-bit NPU architecture and increased memory access for the latest AI models, to accelerate media editing, multimodal content generation, and agentic AI.

[Learn more](https://www.qualcomm.com/processors/hexagon)

### Local AI Performance

Run GenAI workloads locally with low latency and responsive inference.

### Power Efficiency

Sustain AI experiences without excessive power consumption.

### Broad AI Ecosystem

Use the frameworks, runtimes, and models you already know.

### Optimized for AI

Built for Copilot+ PCs and Windows AI workflows.

Hexagon NPU in Snapdragon X2 Elite Extreme delivers on average up to

### 2x AI performance and efficiency vs. competitor GPUs and NPUs

![](https://s7d1.scene7.com/is/image/dmqualcommprod/image-test?$QC_Responsive$&fmt=png-alpha)

Hexagon NPU in the Snapdragon X2 Elite Extreme delivers better AI performance and power efficiency vs. the competition.²

## Case Studies

Check out these applications already using agentic AI on Windows on Snapdragon X Series.

2:08

![LLMWare ModelHQ on Snapdragon](https://s7d1.scene7.com/is/image/dmqualcommprod/LLMware-AVS?req=tmb&wid=640)

### LLMWare ModelHQ on Snapdragon

2:06

![Moises Live](https://s7d1.scene7.com/is/image/dmqualcommprod/6368009779112-1?req=tmb&wid=640)

### Moises Live

1:14

![Realtime instrument isolation in DJ Pro](https://s7d1.scene7.com/is/image/dmqualcommprod/6361210474112-1?req=tmb&wid=640)

### Realtime instrument isolation in DJ Pro

1:02

![Magic Mask in Blackmagic Design’s DaVinci Resolve](https://s7d1.scene7.com/is/image/dmqualcommprod/6361212753112?req=tmb&wid=640)

### Magic Mask in Blackmagic Design’s DaVinci Resolve

8:01

![Run powerful LLMs on NPU with AnythingLLM](https://s7d1.scene7.com/is/image/dmqualcommprod/6368721780112-1?req=tmb&wid=640)

### Run powerful LLMs on NPU with AnythingLLM

1:09

![Enhanced Webcam Quality with Windows Studio Effects](https://s7d1.scene7.com/is/image/dmqualcommprod/6361213777112?req=tmb&wid=640)

### Enhanced Webcam Quality with Windows Studio Effects

4:44

![Port & Optimize Apps for Windows on Snapdragon](https://s7d1.scene7.com/is/image/dmqualcommprod/6397250749112-1?req=tmb&wid=640)

### Port & Optimize Apps for Windows on Snapdragon

## Build with our Trusted Ecosystem Partners

The Hexagon NPU in the Snapdragon X Series delivers fast, efficient on-device AI inference, with an architecture optimized for the latest AI models and use cases, while maintaining low power consumption. Check out our growing ecosystem of partners enabling AI development on PCs with Snapdragon X Series.

[View all models](https://aihub.qualcomm.com/models)

### AI Model Zoos

![](https://s7d1.scene7.com/is/image/dmqualcommprod/ai-hub-4?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/hugging?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/foundry-local?$QC_Responsive$&fmt=png-alpha)

### Models

100s of optimized ecosystem models

Day 0 Support announced for latest models

![](https://s7d1.scene7.com/is/image/dmqualcommprod/meta?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/ibm-1?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/stability?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/mistral?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/gemma?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/deepseek?$QC_Responsive$&fmt=png-alpha)

### AI Scaling Partners

Scale to millions of devs

![](https://s7d1.scene7.com/is/image/dmqualcommprod/ollama?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/llmware?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/unreal?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/lm-studio?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/anything?$QC_Responsive$&fmt=png-alpha)

### AI Frameworks

![](https://s7d1.scene7.com/is/image/dmqualcommprod/onnx?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/llama?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/pytorch-1?$QC_Responsive$&fmt=png-alpha)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/mlc-1?$QC_Responsive$&fmt=png-alpha)

## Evaluate Models and Apps

Discover the power of the Copilot+ PCs powered by Snapdragon X Series on a cloud hosted infrastructure.

Remotely access real devices powered by Snapdragon and Qualcomm® platforms for seamless application evaluations.

### Models

Choose from a variety of models with these options:

#### Qualcomm AI Hub

Browse 300+ AI models optimized and validated by Qualcomm Technologies. Evaluate and optimize models on the Qualcomm AI Hub.

[Learn More](https://aihub.qualcomm.com/)

#### Microsoft Foundry Local

Download AI models optimized for Snapdragon X Series PCs and execute them through the ONNX Runtime on Windows ML.

[Learn More](https://github.com/microsoft/Foundry-Local)

### AI Apps

Clone and customize sample application code for bundling a model to deploy on device.

[Get started with Qualcomm® AI Hub Apps](https://www.qualcomm.com/)

#### Qualcomm AI Hub Workbench

Optimize your custom trained model for a target runtime, run inference. Profile on-device performance with a few lines of code. Target 50+ hosted Qualcomm devices, provisioned within minutes.

[Learn More](https://aihub.qualcomm.com/)

#### Qualcomm Device Cloud (QDC)

Remotely access real devices powered by Snapdragon® and Qualcomm® platforms for seamless testing applications.

[Learn More](https://qdc.qualcomm.com/)

## Build AI Your Way

Flexibility, interoperability and ease-of-use are at the heart of driving innovative uses cases on Snapdragon X Series PCs. Learn how the frameworks and tools of Qualcomm AI provide interoperability with popular model repositories, ecosystem frameworks, and execution providers to develop AI apps on the NPU, CPU, and GPU.

[Windows on Snapdragon AI Documentation](https://docs.qualcomm.com/nav/home/ai-app-development.html?product=1601111740057789)

[View all tools](https://www.qualcomm.com/developer/windows-on-snapdragon/software)

![](https://www.qualcomm.com/content/dam/qcomm-martech/dm-assets/images/developer/wos/redesign/redesign-2025/ai-redesign/diagram-new.svg)

![](https://www.qualcomm.com/content/dam/qcomm-martech/dm-assets/images/developer/wos/redesign/redesign-2025/ai-redesign/diagram-new.svg)

![](https://www.qualcomm.com/content/dam/qcomm-martech/dm-assets/images/developer/wos/redesign/redesign-2025/ai-redesign/diagram-new.svg)

### Choose

Choose from a plethora of foundational models pre-optimized for the Hexagon NPU architecture available from the Microsoft AI Foundry or the Qualcomm AI Hub, or you can bring your own custom models. Build traditional or Generative AI applications on top of popular ecosystem frameworks such as ONNX, PyTorch, TensorFlow and TensorFlow Lite. Choose a runtime framework to meet your development needs and use an Execution Provider listed below.

### Build

Use a guided approach to develop high-performance AI applications with NPU differentiation with Qualcomm AI Engine Direct. The use of ONNX Runtime (ORT) provides scalability, and Windows ML offers IHV interoperability.

### Optimize

Accelerate AI workloads on the CPU, GPU and the NPU engines enabling low latency, high throughput, and low-power usages on the Copilot+ PCs with Snapdragon.

## Execution Providers

### ONNX Runtime

Get the essentials of building AI apps using the ONNX Runtime (ORT) with a flexible interface to integrate hardware-specific libraries (Execution Providers). Take advantage of Olive model optimizations and app-acceleration using QNN-EP for NPU, GPU, CPU, and ORT CPU EP.

[Learn More](https://docs.qualcomm.com/nav/home/ort.html?product=1601111740057789)

### Windows ML

Leverage Windows ML which serves as the AI inferencing core of Windows AI Foundry for AI app development. Using Windows AI APIs and Foundry local models, build apps that seamlessly leverage device capabilities for model inference. Windows ML automatically selects the best hardware, balancing the load and enhancing performance.

[Learn More](https://docs.qualcomm.com/nav/home/winml-ep.html?product=1601111740057789)

### Qualcomm AI Engine Direct (QNN)

Achieve the best possible performance for AI apps on PCs with Snapdragon X Series using Qualcomm AI Engine Direct. The SDK provides tools to support popular AI frameworks and offline tools to convert, quantize, optimize, and offload the models onto various AI Engine runtimes.

[Learn More](https://docs.qualcomm.com/nav/home/qnn.html?product=1601111740057789)

### Generative AI

Leverage GenAI APIs and third-party platforms that offer methodologies, best practices, and tools for deploying generative AI on Windows on Snapdragon devices.

[Learn More](https://docs.qualcomm.com/nav/home/gen-ai-llm.html?product=1601111740057789)

## Samples

Check out these sample apps to learn more about building AI on Windows on Snapdragon.

### [Run DeepSeek Models on Windows on Snapdragon Llama.cpp and MLC-LLM tutorial](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-llama-cpp-mlc-llm)

[Get started](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-llama-cpp-mlc-llm)

### [Run DeepSeek Models on Windows on Snapdragon Ollama tutorial](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-ollama)

[Get started](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-ollama)

### [Enable Pose Detection on Snapdragon X Elite: step-by-step tutorial](https://www.qualcomm.com/developer/blog/2025/03/enable-pose-detection-snapdragon-x-elite-step-by-step-tutorial)

[Get started](https://www.qualcomm.com/developer/blog/2025/03/enable-pose-detection-snapdragon-x-elite-step-by-step-tutorial)

### [Run DeepSeek Models on Windows on Snapdragon LM Studio tutorial](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-lmstudio)

[Get started](https://www.qualcomm.com/developer/blog/2025/02/how-to-run-deepseek-windows-snapdragon-tutorial-lmstudio)

### [Qualcomm® AI Hub sample](https://github.com/quic/ai-hub-models)

[Get started](https://github.com/quic/ai-hub-models)

### [Stable Diffusion sample](https://docs.qualcomm.com/bundle/publicresource/topics/80-64748-1/introduction.html)

[Get started](https://docs.qualcomm.com/bundle/publicresource/topics/80-64748-1/introduction.html)

### [Inceptionv3 demo](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/tutorial3.html)

[Get started](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/tutorial3.html)

### [Mobilenetv2 ONNX Runtime QNN EP Sample](https://github.com/microsoft/onnxruntime-inference-examples/tree/main/c_cxx/QNN_EP/mobilenetv2_classification)

[Get started](https://github.com/microsoft/onnxruntime-inference-examples/tree/main/c_cxx/QNN_EP/mobilenetv2_classification)

![](https://www.qualcomm.com/content/dam/qcomm-martech/dm-assets/images/developer/wos/redesign/redesign-2025/ai-redesign/diagram-new.svg)

## Build Agentic AI apps

![](https://s7d1.scene7.com/is/image/dmqualcommprod/agentic-ai?fmt=webp-alpha&qlt=85)

Build autonomous on-device agents to simplify complex, repetitive, multi-step, schedule-driven enterprise workflows. Take advantage of PCs with Snapdragon X Series to transform enterprise productivity with fast, private, local agentic AI execution for real-world use cases.

[Learn more](https://www.qualcomm.com/developer/blog/2026/05/local-agentic-ai-with-llmware-on-pcs-with-snapdragon-x-series)

## Discover our latest insights

Developer Blog

### [How to write fast AArch64 SVE/SME Kernels with Ripple and LLVM](https://www.qualcomm.com/developer/blog/2026/10/ripple-llvm-aarch64-sve-sme-kernel-implementation)

Oct 29, 2026

[LLVM](https://www.qualcomm.com/developer/blog?tags=LLVM) [Open Source](https://www.qualcomm.com/developer/blog?tags=Open%20Source)

Developer Blog

### [Same-day upstream Linux support now posted for Snapdragon 8 Elite Gen 6 and Snapdragon Elite Extreme Gen 6 mobile platforms](https://www.qualcomm.com/developer/blog/2026/10/same-day-snapdragon-8-elite-gen-6-upstream-linux-support)

Oct 27, 2026

[Adreno](https://www.qualcomm.com/developer/blog?tags=Adreno) [Linux](https://www.qualcomm.com/developer/blog?tags=Linux) [Snapdragon](https://www.qualcomm.com/developer/blog?tags=Snapdragon)

Developer Blog

### [How to share Vulkan and OpenCL workloads on Android with Qualcomm Adreno GPUs](https://www.qualcomm.com/developer/blog/2026/10/how-to-share-vulkan-opencl-workloads-android-adreno-gpus)

Oct 22, 2026

[Vulkan](https://www.qualcomm.com/developer/blog?tags=Vulkan) [OpenCL](https://www.qualcomm.com/developer/blog?tags=OpenCL) [Android](https://www.qualcomm.com/developer/blog?tags=Android)

Developer Blog

### [Port x64 Windows projects to Windows on Snapdragon using agentic AI](https://www.qualcomm.com/developer/blog/2026/10/port-windows-projects-to-windows-on-snapdragon-with-agentic-ai)

Oct 20, 2026

[Open Source](https://www.qualcomm.com/developer/blog?tags=Open%20Source) [AI](https://www.qualcomm.com/developer/blog?tags=AI) [Windows on Snapdragon](https://www.qualcomm.com/developer/blog?tags=Windows%20on%20Snapdragon)

Developer Blog

### [Native Python on Windows on Snapdragon: a stronger foundation for developers and AI workflows](https://www.qualcomm.com/developer/blog/2026/10/native-python-windows-on-snapdragon-stronger-foundation-for-developers-ai-workflows)

Oct 14, 2026

[Windows on Snapdragon](https://www.qualcomm.com/developer/blog?tags=Windows%20on%20Snapdragon)

## Connect with our communities

[LinkedIn](https://www.linkedin.com/showcase/qualcomm-developer)

[See what's new](https://x.com/Qualcomm_Dev)

[YouTube](https://www.youtube.com/qualcommdev)

[GitHub](https://github.com/qualcomm)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/discord-icon-1?$QC_Responsive$&fmt=png-alpha)

[Discord](https://discord.com/invite/qualcommdevelopernetwork)

[Visit the forum](https://mysupport.qualcomm.com/supportforums/s/)

![](https://s7d1.scene7.com/is/image/dmqualcommprod/ai-hub-bg-desktop?fmt=webp-alpha&qlt=85)

¹ Features: [Snapdragon X2 Elite](https://www.qualcomm.com/laptops/products/snapdragon-x2-elite)

² Performance & power efficiency is based on UL Procyon AI Computer Vision score on Windows 11 OS, run in September 2025 and April 2026. Snapdragon X2 Elite Extreme (X2E-96-100) was tested using a Qualcomm reference design. Snapdragon X Elite (X2E-96-100) was tested using a Qualcomm reference design Intel Core Ultra X7 358H was tested using an HP OmniBook X 16" (16-BG0013DX) Intel Core Ultra 9 285H was tested using an Asus ZenBook 14 (UX3405CA). Intel Core Ultra 9 288V was tested using an Acer Swift 16 AI OLED (SF16-51T-98SE). Power and performance comparison reflect results based on measurements and hardware instrumentation of given devices.

- Best-in-class Qualcomm Oryon™
- CPU up to 18 cores for the fastest and most power-efficient performance at up to 5.0 GHz frequency
- Up to 1.85 GHz GPU
- Multitasking across up to three 5K 60Hz monitors
- Multi-day battery life
- Truly mobile PC with sustained performance when on battery
- Optimized power consumption spreads workloads across the CPU, GPU, and NPU
- Featuring up to 80 TOPS of AI processing
- AI acceleration for media editing, multimodal content generation, and agentic AI
- 64-bit NPU architecture with increased memory access to the latest AI models
- Powers Copilot+ PCs for the fastest, most intelligent Windows PCs
- Lightning-fast 5G for up to 10 Gbps peak speeds
- Wi-Fi 7 with High Band Simultaneous for peak speeds of 5.8 Gbps
- Dual Bluetooth with Snapdragon Sound support