⚡ GenieX: Cross-Platform On-Device AI Runtime by Qualcomm

Open-sourced by Qualcomm, this cross-platform AI runtime is designed to run the latest large language models (LLMs) and vision-language models (VLMs) efficiently across various hardware. It supports GPUs and CPUs, with a particular emphasis on NPU utilization. The scope is broad, covering runtime environments from PCs (Python/C++) and mobile devices (Android/iOS) to Linux/IoT setups. A notable feature is its commitment to "day-0" support for new models, already boasting compatibility with frontier models like Qwen-3-VL and Gemma-3n. For developers needing to deploy AI capabilities directly on edge devices, this is a substantial SDK. Backed by a major chip manufacturer, it aims to provide a unified and efficient inference interface between underlying hardware and upper-level models.