• Powerful CPU+GPU+NPU Heterogeneous Computing Aurora AI1000 delivers heterogeneous computing power up to 50 TOPS (INT8) and 12 TFLOPS (FP16). Built around an 8-core high-performance SoC running at 2.65GHz with integrated GPU (≥3 TFLOPS) and programmable NPU, it handles multimodal AI tasks with low latency for local inference.
• Ultra-Compact 0.7L Mini Chassis, Desktop-Grade AI Performance Measuring only 127×127×44 mm, this 0.7L edge AI box is smaller than a dictionary while maintaining desktop-level AI productivity. Equipped with active cooling, it runs stably for long-duration AI workloads and is easy to deploy in limited-space industrial and office environments.
• Ready for Local Small-Scale AI Model Deployment Pre-compatible with mainstream open-source AI models including Whisper, Piper TTS and SadTalker. Supports local small model fine-tuning, prototype verification and offline inference, protecting sensitive data privacy and cutting enterprise AI development costs.
• Full-Featured Hardware & Rich Connectivity Comes with 32GB LPDDR5/5X memory and 1TB NVMe M.2 SSD. Equipped with dual Gigabit Ethernet, Wi‑Fi 6, Bluetooth, 2×HDMI®1.4, 5×USB3.0-A, 1×USB-C and 3.5mm audio jack. Multiple ports support multi-screen output and connection to various sensors and peripherals.
• Wide Versatile AI Application Scenarios Perfect for AI digital human, industrial robots, factory automation, local voice assistant and edge AI inference. It also works as a robot controller, IoT industrial gateway, programming learning workstation, lightweight content creation and compact gaming machine.

The Aurora AI1000 is a 0.7L ultra-compact edge AI box that brings desktop-level AI productivity in a dictionary-sized form factor. Powered by CPU+GPU+NPU heterogeneous architecture, it achieves 50 TOPS INT8 and 12 TFLOPS FP16 computing power, built on an 8-core 2.65GHz high-performance SoC.

It supports local deployment of small AI models and is pre-adapted for Whisper, Piper TTS, SadTalker and other popular AI projects. With 32GB LPDDR5/5X RAM and 1TB NVMe M.2 SSD, paired with active cooling and comprehensive interfaces including Wi-Fi6, dual Gigabit LAN and dual HDMI outputs, it supports multi-screen display and abundant peripheral access.

Running on Ubuntu system, this compact device serves as edge AI node for industrial automation, AI digital human, robot control and offline local LLM inference. It is an ideal low-cost platform for AI prototype testing, education research, IoT gateway, lightweight content creation and home entertainment.