Partnerships
Ceva NeuPro NPU Family Reaches Production in Mobilint REGULUS Edge AI SoC
Add Unite.AI to your preferred sources on Google
Ceva announced on November 2, 2026 that Mobilint has deployed Ceva’s SensPro AI DSP and NeuPro-Nano NPU in REGULUS, Mobilint’s edge AI system-on-chip, in what Ceva describes as the first production deployment of its NeuPro NPU family.
Mobilint is a fabless AI semiconductor company founded in 2019 and headquartered in Seoul, specializing in NPUs and AI SoCs for edge and on-device AI applications. Ceva licenses silicon and software IP for smart edge devices and says its technology has shipped in more than 22 billion devices for over 400 customers.
Inside the REGULUS SoC
Designed as a standalone processor for on-device AI, REGULUS integrates the SensPro AI DSP and the NeuPro-Nano NPN32 NPU within Mobilint’s processing architecture. According to the announcement, the chip delivers 10 TOPS of AI performance at under 3W and supports up to 8GB of memory, with Mobilint’s REGULUS product page specifying dense INT8 compute on a 17mm chip. The design targets real-time AI in power- and space-constrained applications including robotics, drones, machine vision, smart cameras and IoT devices.
The multi-engine architecture assigns different sensing and AI workloads to the processing engine best suited to each task. SensPro provides programmable processing for sensing and signal-processing workloads, while NeuPro-Nano handles neural processing for always-on inference use cases, complementing Mobilint’s proprietary AI processing architecture.
Beyond inference, the SoC includes a built-in image processor with HDR and noise reduction, hardware video encoding at 4K 60 frames per second, and direct connections for cameras, displays, Ethernet, USB and sensors. Mobilint positions REGULUS as a complete processor that runs Linux and applications without a host, rather than as an accelerator.
On the software side, Mobilint’s SDK qb compiles PyTorch, TensorFlow and ONNX models for the chip, and the company says more than 490 model architectures across CNN, RNN and transformer families have been validated on its hardware. A Yocto Linux SDK includes GStreamer and OpenCV, with camera, video and display drivers provided.
MLD-R1 and the REGULUS Product Family
REGULUS already powers Mobilint’s MLD-R1 AI USB accelerator, first previewed at COMPUTEX 2026 and now shipping. Designed to add on-device AI to existing systems, the MLD-R1 supports computer vision, video analytics and generative AI workloads for applications including smart factories, robotics, machine vision, security, medical devices and digital signage, with support for the PyTorch, ONNX, Hugging Face, TensorFlow, TFLite and Keras frameworks.
“REGULUS brings together the specialized processing capabilities needed to run sophisticated AI workloads directly on the device within very tight power constraints,” said Dongjoo Shin, CEO of Mobilint. Shin said the Ceva IP adds flexible sensing, signal processing and inference capabilities for the workloads customers need to build the next generation of intelligent edge devices.
Mobilint offers REGULUS in three forms: the MLM-1 system-on-module for adding AI to custom carrier boards, the REGULUS AI EVK evaluation kit for benchmarking and software bring-up, and the MLD-R1 USB-C dongle for adding AI to existing PCs. The company lists target devices including AI cameras for production-line inspection, drones, AI CCTV, robots, kiosks and aerospace platforms.
Yaron Galitzky, executive vice president and general manager of Ceva’s AI division, said REGULUS “marks an important milestone for Ceva as the first production deployment of our NeuPro NPU family,” adding that NeuPro-Nano and SensPro together provide efficient, flexible sensing and inference processing for the workloads physical AI devices need to perform locally and in real time.
