Edge AI Compute Platforms
A formidable Agentic OS requires relentless hardware. Forged from autonomous-driving DNA, our AI Boxes deliver scalable edge inference built for the physical world — from agile deployment to massive industrial throughput.
Ultra-Compact · Custom-Built
A highly compact metal chassis with dual-fan thermal architecture, designed to embed seamlessly into third-party bodies. We offer deep hardware-level customization — heterogeneous form-factor adaptation, specialized industrial I/O expansion, and ruggedized protection for extreme environments. We build it to fit your exact product definition.
From 80 TOPS and 128 TOPS up to a staggering 560+ TOPS — breaking the edge compute bottleneck.
Multi-core CPUs deliver ultra-low scheduling latency, keeping Model Predictive Control loops under 20 ms.
Native GMSL for auto-grade HD video, up to 8× USB-C — fully compatible with industrial depth cameras and LiDAR.
Multiple GbE ports and built-in CAN buses interface directly with chassis and servos, plus high-speed Wi-Fi and UFS/SSD storage.
Frontier Policies, Running on the Box
State-of-the-art VLA models — smolVLA and π0 — deployed and inferring live on a single iRC100P, from a terminal command to real-world manipulation.
A compact vision-language-action policy driving real-time pick-and-place — running fully on-device.
The π0 foundation policy deployed at the edge — from a single command in the terminal to live manipulation.