# Compile and optimize an AI model

Use one of the following options to compile and optimize your model.

<svg xmlns="http://www.w3.org/2000/svg" width="16" height="17" viewbox="0 0 16 17" fill="none" aria-label="icon-book">
<path d="M8 2.5V14.5M3.33333 2.5H12.6667C13.403 2.5 14 3.09695 14 3.83333V13.1667C14 13.903 13.403 14.5 12.6667 14.5H3.33333C2.59695 14.5 2 13.903 2 13.1667V3.83333C2 3.09695 2.59695 2.5 3.33333 2.5Z" stroke="#717171" stroke-width="1.33333" stroke-linecap="round" stroke-linejoin="round"></path>
</svg> Use AI Hub to optimize an AI model

Use AI Hub to download a preoptimized model or optimize your own model.

https://docs.qualcomm.com/doc/80-80022-15B/topic/ai-hub.html

<svg xmlns="http://www.w3.org/2000/svg" width="16" height="17" viewbox="0 0 16 17" fill="none" aria-label="icon-book">
<path d="M8 2.5V14.5M3.33333 2.5H12.6667C13.403 2.5 14 3.09695 14 3.83333V13.1667C14 13.903 13.403 14.5 12.6667 14.5H3.33333C2.59695 14.5 2 13.903 2 13.1667V3.83333C2 3.09695 2.59695 2.5 3.33333 2.5Z" stroke="#717171" stroke-width="1.33333" stroke-linecap="round" stroke-linejoin="round"></path>
</svg> Use Qualcomm AI Runtime SDK to optimize an AI model

Port your models using the Qualcomm AI Runtime (QAIRT) SDK.

https://docs.qualcomm.com/doc/80-80022-15B/topic/qairt.html

<svg xmlns="http://www.w3.org/2000/svg" width="16" height="17" viewbox="0 0 16 17" fill="none" aria-label="icon-book">
<path d="M8 2.5V14.5M3.33333 2.5H12.6667C13.403 2.5 14 3.09695 14 3.83333V13.1667C14 13.903 13.403 14.5 12.6667 14.5H3.33333C2.59695 14.5 2 13.903 2 13.1667V3.83333C2 3.09695 2.59695 2.5 3.33333 2.5Z" stroke="#717171" stroke-width="1.33333" stroke-linecap="round" stroke-linejoin="round"></path>
</svg> Fine tune an AI model with custom data using Edge Impulse

Fine tune models using your own data and Edge Impulse.

https://docs.qualcomm.com/doc/80-80022-15B/topic/edge-impulse.html

<svg xmlns="http://www.w3.org/2000/svg" width="16" height="17" viewbox="0 0 16 17" fill="none" aria-label="icon-book">
<path d="M8 2.5V14.5M3.33333 2.5H12.6667C13.403 2.5 14 3.09695 14 3.83333V13.1667C14 13.903 13.403 14.5 12.6667 14.5H3.33333C2.59695 14.5 2 13.903 2 13.1667V3.83333C2 3.09695 2.59695 2.5 3.33333 2.5Z" stroke="#717171" stroke-width="1.33333" stroke-linecap="round" stroke-linejoin="round"></path>
</svg> Fine tune model accuracy using Qualcomm AI Model Efficiency Toolkit

Fine tune model accuracy using the Qualcomm AI Model Efficiency Toolkit.

https://docs.qualcomm.com/doc/80-80022-15B/topic/aimet.html

Last Published: May 14, 2026

[Previous Topic
Run a sample application using Qualcomm Device Cloud](https://docs.qualcomm.com/bundle/publicresource/80-80022-15B/topics/sample-app-qdc.md) [Next Topic
Use AI Hub to optimize a model](https://docs.qualcomm.com/bundle/publicresource/80-80022-15B/topics/ai-hub.md)