GenieX

GenieX is an on-device generative AI inference runtime designed specifically for Qualcomm Snapdragon platforms. It enables local execution of advanced language models (LLMs) and vision-language models (VLMs), allowing developers to run these models directly on device hardware such as the Hexagon NPU, Adreno GPU, or CPU. GenieX is positioned as the community version of Qualcomm GENIE and is currently available in a developer preview.

The platform provides multiple entry points for integrating AI inference into applications. Developers can interact with GenieX through a command-line interface for running and serving models directly from the terminal, a Python SDK for embedding inference capabilities into Python applications, and a Java/Kotlin SDK tailored for on-device mobile apps on Android. Additionally, GenieX offers a Docker container for reproducible deployments and an OpenAI-compatible local server, facilitating integration with existing OpenAI-compatible workflows.

GenieX is intended for developers seeking to deploy frontier generative AI models on Qualcomm hardware across a variety of environments, including mobile devices and embedded systems. Its architecture is designed to simplify the process of running sophisticated AI models locally, requiring only a few lines of code to get started.

evidence_sufficient": true}