Qualcomm acquires Nexa AI, open-sources GenAI runtime for Hexagon NPUs (github.com)

Qualcomm has acquired Nexa AI and announced the open-source release of GenieX, an on-device generative AI inference runtime specifically designed for Qualcomm devices. GenieX supports local execution of nearly any GGUF model from platforms like Hugging Face, utilizing Snapdragon’s Hexagon NPUs, Adreno GPUs, and CPUs. The runtime is built with a C SDK, offering interfaces in CLI, Python, Kotlin/Java, Docker, and as an OpenAI-compatible server, allowing developers to engage with a wide array of generative models seamlessly. This development is significant for the AI/ML community as it enhances accessibility and performance for deploying AI models directly on mobile and IoT devices powered by Snapdragon chips. The dual runtime capacity accommodates both models sourced from Hugging Face and pre-compiled bundles from Qualcomm AI Hub, providing flexibility for developers to leverage existing models or optimize performance specifically for Qualcomm hardware. With comprehensive documentation and support for various programming environments, GenieX is positioned to accelerate the integration of generative AI into consumer devices while inviting contributions from the developer community.