Arduino turned 21 and shipped its first real edge-AI SBC. The VENTUNO Q pairs a Qualcomm Dragonwing IQ-8275 (40 TOPS NPU, 8-core Kryo, Adreno GPU) with an STM32H5 real-time MCU on one board, bridged over RPC. Under $300, Q2 2026 via Arduino Store, DigiKey, Mouser, Farnell.
Hardware, not hype
This is a single-board computer built for robots, not a Pi clone. 16 GB RAM, 64 GB expandable storage, M.2 NVMe Gen 4, three MIPI-CSI camera ports for 360° vision, Wi-Fi 6, BT 5.3, 2.5 Gb Ethernet, CAN-FD. The IQ8 runs Ubuntu or Debian with ROS 2 for perception and planning; the STM32H5 Cortex-M33 at 250 MHz handles deterministic motor control and sensing. One board does perception, decision, and actuation — the thing Jetson Orin Nano hobbyists have been stitching together with a second MCU for years.
How an agent actually drives it
Arduino App Lab is the entry point. It ships a curated catalog of offline models — local LLMs, VLMs, ASR, gesture, pose, object tracking — plus full Edge Impulse Studio integration for custom training. Flash it from Arduino sketches, Python scripts, or pull pre-optimized models from Qualcomm AI Hub. The RPC bridge is the interesting bit: an agent running a VLM on the IQ8 side can call into the STM32 to drive a servo arm without leaving the board. No cloud. No latency tax.
You Might Also Like
- Orange pi 6 Plus a 199 Single Board Computer With 45 Tops and 64gb ram
- Olares one April Shipping Batch Ships rtx 5090 Mobile 96gb ram for 70b Local Llms on Your Desk
- Huawei ai Glasses Launch April 21 11 Hour Battery Built in Camera and a Harmonyos sdk Play
- Google and Synaptics put two Npus on one Edge ai Board the Synaptics Coral dev Board Ships With Gemma 3 Ready to go
- Megatrain Trains 100b Parameter Llms on a Single gpu With 1 5tb of ram
Top AI Product 
Leave a comment