Sixfab AI HAT+ brings 25 TOPS NPU acceleration to Raspberry Pi 5

Sixfab and DEEPX have introduced the AI HAT+ for Raspberry Pi 5, a third-party accelerator designed for edge AI workloads. Built around the DEEPX DX-M1M NPU, the board provides 25 TOPS of dedicated performance while consuming approximately 3W of sustained power. This allows the Raspberry Pi CPU to focus on application logic and connectivity while the NPU handles CNN, VLM, and SLM tasks locally.

Technical capabilities of the DX-M1M NPU

• Performance: Delivers 25 TOPS of AI acceleration.
• Power Efficiency: Typical sustained NPU power consumption is approximately 3W.
• Memory: Includes 2 GB of dedicated on-module LPDDR4x memory.
• Workload Support: Optimized for Convolutional Neural Networks (CNN), compact Vision-Language Models (VLM), and Small Language Models (SLM).
• Software Stack: Utilizes DX-COM for model optimization, DX-RT for C++/Python APIs, and DX-Stream for GStreamer-based pipelines.

Why it matters

The accelerator enables low-power, privacy-focused local AI on Raspberry Pi hardware, reducing reliance on cloud connectivity for vision and language tasks.