Support and discussions for the Sixfab AI HAT+: build powerful edge AI applications on Raspberry Pi 5 with fast, on-device inference using the DEEPX-powered Sixfab AI accelerator.
Welcome to the AI HAT+ category!
This is your dedicated community space for exploring what’s possible with the Sixfab AI HAT+ for Raspberry Pi 5, from your very first AI demo to fully deployed, real-world edge AI systems. Explore how other developers are using the board in creative and practical ways, learn from shared experiences, and get guidance for your own projects.
What is the Sixfab AI HAT+?
The Sixfab AI HAT+ is a high-performance AI accelerator for Raspberry Pi 5 with a dedicated DEEPX NPU, enabling real-time AI inference on-device for edge AI and emerging physical AI applications. It brings high-performance inference directly to your Raspberry Pi, enabling fast, private, and reliable AI applications without depending on the cloud.
Whether you’re building an intelligent camera that detects objects in real time, developing a robotics project with visual awareness, or creating a smart IoT gateway that makes decisions locally, this is where your ideas come to life.
Why PCIe instead of a USB AI accelerator?
Thanks to its PCIe Gen 3 x1 connection, the AI HAT+ offers significantly higher speed and lower latency compared to traditional USB-based AI accelerators. This makes it ideal for demanding workloads like video analytics and multi-stream processing on Raspberry Pi 5.
The Sixfab AI HAT+ enables:
- Real-time AI inference at the edge
- Local data processing with improved privacy
- High-performance edge computing without cloud dependency
Which variant should I choose: 13 TOPS or 25 TOPS?
| AI HAT+ 13 TOPS | AI HAT+ 25 TOPS | |
|---|---|---|
| NPU | DEEPX DX-M1ML | DEEPX DX-M1M |
| Compute (INT8) | 13 TOPS | 25 TOPS |
| NPU memory | 1 GB LPDDR4x | 2 GB LPDDR4x |
| Best for | Single-stream vision, prototypes, budget-sensitive projects | Multi-stream 1080p, larger models, production deployments |
Both variants share the same PCB, HAT+ form factor, and software stack; only the NPU changes.
Topics covered in this category
- Getting started with Sixfab AI HAT+ and Raspberry Pi 5
- Setting up and running your first AI models on Sixfab AI HAT+
- Using the Sixfab Model Zoo for rapid prototyping
- Building custom pipelines with the DXNN SDK on AI HAT+
- Optimizing performance for real-time inference
- Multi-camera and multi-stream AI applications
- Thermal management and hardware optimization
- Integrating cameras, sensors, and IoT systems
- Comparing AI HAT+ 13 TOPS and 25 TOPS in real-world scenarios
- Debugging, troubleshooting, and best practices
What you can build
- Real-time object detection and smart security systems
- AI-powered robotics with vision and decision-making
- Smart city and traffic monitoring solutions
- Industrial inspection and automation systems
- Intelligent IoT gateways with local AI processing
- Audio and video analytics at the edge
From quick prototypes using ready-to-run models to fully customized AI pipelines, the Sixfab AI HAT+ supports you at every stage of development. Join the conversation to share your projects, ask questions, exchange ideas, and push the boundaries of what edge AI can do in real-world environments.
Products and resources
| Resource | Link |
|---|---|
 Buy AI HAT+ |
sixfab.com/product/ai-hat-plus-raspberry-pi-5 |
 Documentation |
AI HAT+ Overview |
 Quickstart |
AI HAT+ Quickstart |
 Troubleshooting & FAQ |
Troubleshooting · FAQ |
 Model deployment |
Sixfab Model Zoo · DXNN SDK Workflow |
 DEEPX examples on GitHub |
github.com/sixfab/sixfab-dx-examples |
 All Raspberry Pi HATs |
sixfab.com/product-category/development-tools/raspberry-pi-hats |