Mobilint said on Nov. 4 that it has improved support for its neural processing units (NPUs) in Ray, enabling the software framework to automatically detect its chips. Ray is an open-source software framework designed to efficiently handle large-scale artificial intelligence (AI) workloads, connecting multiple servers and AI chips so their computing resources can be used collectively. Mobilint worked with the Electronics and Telecommunications Research Institute (ETRI) to develop a feature that enables Ray to automatically detect its NPUs and allocate computing resources; previously, developers had to go through complex procedures, including manually registering the processors, to use Mobilint NPUs with Ray. The feature underwent an official code review by the Ray project and was incorporated into its official GitHub repository in August 2026. Mobilint NPU users can now more easily build and operate AI systems using multiple NPUs through Ray. “We have expanded the foundation for Mobilint NPUs to be used in the global AI software ecosystem,” the company said, adding that beyond computing performance and power efficiency of AI chips themselves, software compatibility and ease of development are increasingly becoming important competitive factors. Mobilint added that it is enhancing its proprietary software development kit (SDK) and supporting software technologies, and is expanding its presence in the AI market with its high-performance AI accelerator ARIES and on-device AI system-on-chip REGULUS. REGULUS incorporates a video processing unit (VPU) from Chips&Media and has also completed performance validation in