Raspberry Pi AI HAT+ 2 40 TOPS
Raspberry Pi AI HAT+ 2（最大40 TOPS）樹莓派原廠人工智慧擴充板，內建Hailo-10H、Pi5擴展板
*樹莓派原廠人工智慧擴充板，內建Hailo-10H AI加速器、快速建構各種AI驅動的應用程式
*Raspberry Pi AI HAT+ 2 樹莓派人工智慧套件 PCIe 擴充套件 樹莓派5 Raspberry Pi5
提供40TOPS版本，相容於Pi5 PCIe Gen 3標準，符合HAT+標準
Raspberry Pi AI HAT+ 2 is an add-on board based on the Hailo-10H AI accelerator that brings generative AI capability to Raspberry Pi 5.
〔Product includes〕清單：
- Raspberry Pi AI HAT+ 2
- 16mm Stacking Header
- Spacers and Screws
［產品特性］
Building on the success of the original AI HAT+, it has been fully upgraded to efficiently accelerate selected Large Language Models (LLMs) and Vision-Language Models (VLMs), specifically designed for Raspberry Pi 5. It offers an accessible, cost-effective, and energy-efficient way to integrate high-performance artificial intelligence technology. Widely applicable to scenarios such as intelligent monitoring, smart home control, robotics development, industrial process inspection, and intelligent interactive terminals.
［Features & Specs］主要功能特性：
- Hailo-10H AI accelerator
- 40 TOPS INT4 inference performance
- 8GB on-board RAM
- Generative AI (LLM and VLM) support
- Computer vision performance comparable to AI HAT+
- PCIe interface for Raspberry Pi 5
- Raspberry Pi OS auto-detection／作業系統自動偵測
- rpicam-apps camera pipeline integration
- Pre-trained and custom model support
- Raspberry Pi HAT+ compliant
- Optional heatsink for sustained workloads
- 0°C to 50°C operating temperature
- Mounting hardware included
- Production lifetime through January 2036
【Why the "2" Matters?】
While the original AI HAT+ was designed primarily for computer vision, the AI HAT+ 2 is a significant evolution tailored for the next generation of AI:
＊〔Significant Performance Boost〕：
Jumping to 40 TOPS significantly expands your generative capabilities. This enables the board to run more complex models, such as speech-to-text and live translation, that previously required cloud processing.
＊〔Dedicated 8GB RAM〕：
By having 8GB of local LPDDR4 memory, the HAT+ 2 can load Large Language Models (LLMs) and Vision-Language Models entirely on the card. This eliminates the need to shuttle data back and forth, preventing bottlenecks and leaving the Raspberry Pi 5's system RAM fully available for your OS and other applications.
＊〔True Edge Independence〕：
The combination of higher TOPS and local memory makes this board ideal for offline inference, enabling you to run sophisticated chatbots and visual analysis tools in environments without internet connectivity.
【Seamless Integration】
Designed exclusively for the Raspberry Pi 5, the HAT+ 2 uses the Pi’s PCIe interface for ultra-low-latency, high-bandwidth communication.
＊〔Plug-and-Play〕即插即用：
The Raspberry Pi AI HAT+ 2 is automatically detected by Raspberry Pi OS, requiring minimal configuration to get started.
＊〔Camera Ready〕：
This HAT integrates directly with rpicam-apps, allowing your existing camera pipelines to instantly take advantage of hardware-accelerated post-processing.
【Privacy & Reliability】
By processing data locally rather than sending it to the cloud, the AI HAT+ 2 delivers deterministic, low-latency performance and full data privacy. It is the ideal solution for robotics, industrial monitoring, and secure facility management where internet connectivity is unreliable or security is paramount.
〔Heatsink application instructions〕散熱片安裝說明：
*While the heatsink supplied is optional, users running benchmarks or very intensive AI workloads are advised to install it to prevent thermal throttling.
*雖然提供的散熱片是可選的，但建議運行基準測試或高強度 AI 工作負載的用戶安裝散熱片，以防止過熱降頻。
１． Remove the backing paper from the underside of the product.
２． Make sure your Raspberry Pi AI HAT+ 2 is powered off. Position the heatsink carefully in the correct space on the AI HAT+ 2, making sure not to hit any of the connectors.
３． Align the two black push pins with the two dedicated heatsink holes.
４． When correctly positioned, press evenly on the tops of the two push pins simultaneously until they click, indicating that they are clipped onto the board.
５． We recommend that the heatsink is not removed once it is fitted to the Raspberry Pi AI HAT+ 2.
Removal of the heatsink will cause the push pins and thermal pads to degrade and is likely to lead to product damage.
６． Ensure the push pins are undamaged and securely clipped onto the AI HAT+ 2 before use
