Boardcon Tiny1126B Squeezes AI Vision Into a 34×30mm System-on-Module
Boardcon's new Tiny1126B shrinks its Rockchip RV1126B AI vision module to 34×30mm while adding USB 3.0 and more I/Os — targeting smart cameras, robots, and compact edge AI deployments.
Smaller Footprint, More Capability: Boardcon Tiny1126B
Boardcon announced the Tiny1126B on April 17, 2026 — a compact system-on-module that takes an already space-efficient design and makes it smaller, while simultaneously adding more connectivity. The result is a 34×30mm AI vision module built on the Rockchip RV1126B processor that is positioned for the next generation of compact embedded AI deployments.
The Tiny1126B shrinks Boardcon's previous MINI1126B-P from 38×30mm to 34×30mm — a seemingly minor dimensional reduction that opens up meaningfully tighter enclosures and more constrained installation environments in smart cameras, robotics platforms, and IoT edge devices.
Rockchip RV1126B: The AI Vision Processor
The Rockchip RV1126B is a quad-core Cortex-A53 processor clocked at 1.6GHz, paired with a 3 TOPS neural processing unit (NPU) that supports INT4, INT8, INT16, and FP16 inference formats. The NPU is compatible with TensorFlow, ONNX, PyTorch, and Caffe frameworks — covering the major model training ecosystems without requiring format conversion overhead.
Three TOPS of INT8 inference throughput covers the practical range of edge AI vision tasks: real-time object detection, face recognition, license plate identification, human pose estimation, and driver monitoring system (DMS) inference. For applications requiring visual AI processing without a connection to cloud infrastructure, the RV1126B delivers the capability in a thermally manageable, power-efficient package.
Memory configurations reach up to 4GB of RAM with up to 256GB of eMMC or UFS storage — sufficient for running AI inference pipelines and local data buffering without external storage dependencies.
More I/Os in Less Space
The headline design achievement of the Tiny1126B is the combination of dimensional reduction with expanded connectivity. Despite being smaller than its predecessor, the module switches from two 0.5mm pitch 80-pin connectors to two 0.4mm pitch 100-pin connectors — increasing I/O density as the physical footprint decreases.
New and Expanded Interfaces
The Tiny1126B adds connectivity that the previous module lacked:
- **USB 3.0 DRD (Dual-Role Device)**: Host and device mode support at SuperSpeed bandwidth
- **Additional SDMMC interface**: More flexible storage expansion options
- **Extra SPI and I2C interfaces**: Expanded sensor and peripheral connectivity
- **Fast Ethernet**: Wired network connectivity for industrial and commercial deployments
Target Applications
Boardcon positions the Tiny1126B across several verticals where compact, reliable AI vision processing has a clear product-market fit:
- **Smart cameras and inspection systems**: AI-powered quality control and security in minimal physical footprints
- **Smart door locks and access control**: On-device face recognition with audit trail capabilities
- **Driver Monitoring Systems (DMS) and Blind Spot Detection (BSD)**: Automotive-adjacent safety applications
- **Cleaning and logistics robots**: On-device vision guidance without cloud latency dependency
- **Smart displays**: Interactive AI vision in kiosk and signage form factors
Software and Development Support
Boardcon provides Debian 12 via a Buildroot build system with Linux kernel 6.1.141 and U-Boot 2017.09 — a stable, well-maintained software baseline that simplifies the transition from development to production deployment. Pricing has not been announced at time of writing.
For embedded AI developers evaluating compact vision modules, the Tiny1126B adds a compelling option at the compact end of the RV1126B ecosystem.
Sources: CNX Software (April 17, 2026), Boardcon.com product page (April 2026), CNX Software RV1126B-P module coverage (September 2025)