An AI HAT that understands your workload.
The Raspberry Pi AI HAT+ 2 is a next-generation AI accelerator designed to unlock generative AI on Raspberry Pi 5. Powered by the Hailo-10H AI accelerator and equipped with 8GB of dedicated on-board memory, it enables advanced AI workloads that were previously out of reach on compact edge devices.
By handling AI processing independently, the AI HAT+ 2 allows the Raspberry Pi 5 to remain responsive and focused on system control, networking, and application logic. This architecture makes it especially well suited for large language models (LLMs) and vision-language models (VLMs) running locally, smoothly, and securely.
Designed for privacy-first edge AI, the AI HAT+ 2 delivers fast, low-latency inference without relying on cloud connectivity. Running models directly on the device reduces data exposure, lowers infrastructure complexity, and eliminates recurring cloud API costs — making it ideal for commercial, industrial, and research deployments.
To accelerate development, pre-trained reference models from Hailo are available out of the box. Developers can also train custom vision models or fine-tune generative AI workloads for applications such as speech-to-text, language translation, and visual scene understanding. Software support for compatible generative AI models is maintained through Hailo’s official GitHub repository, ensuring ongoing updates and ecosystem support.
