{"id":348664,"date":"2026-01-15T09:21:07","date_gmt":"2026-01-15T08:21:07","guid":{"rendered":"https:\/\/raspberrypi.dk\/produkt\/raspberry-pi-ai-hat-2\/"},"modified":"2026-03-31T21:05:56","modified_gmt":"2026-03-31T19:05:56","slug":"raspberry-pi-ai-hat-2","status":"publish","type":"product","link":"https:\/\/raspberrypi.dk\/en\/product\/raspberry-pi-ai-hat-2\/","title":{"rendered":"Raspberry Pi AI HAT+ 2"},"content":{"rendered":"
\n
\n
\n

Off\u2011board tasks with on\u2011board RAM<\/h2>\n

With 8GB of dedicated on\u2011board RAM, the AI\u00a0HAT+\u00a02 is ideal for running large language models (LLMs) and vision-language models (VLMs) locally and securely, leaving the host Raspberry\u00a0Pi\u00a05 free to handle other tasks and ensuring your AI models run\u00a0smoothly.<\/p>\n<\/div>\n<\/section>\n

\n
\n
\n
\"\"<\/div>\n
<\/div>\n\"\"<\/picture>\n<\/div>\n<\/div>\n
\n

A model for privacy and security<\/h2>\n

Delivering reliable, low-latency AI processing at the edge and without a network connection, the AI\u00a0HAT+\u00a02 simplifies data security, minimises infrastructure requirements, and reduces cloud API\u00a0costs.<\/p>\n<\/div>\n<\/section>\n<\/div>\n

\n
\n
\n

Ready-made models<\/h2>\n

A set of sample models is provided by Hailo; users can also train custom vision models or fine-tune generative AI models to suit their application, such as speech to text, translation, or visual\u00a0scene\u00a0analysis.<\/p>\n

All of the software for compatible generative AI models can be found in Hailo\u2019s GitHub\u00a0repo.<\/p>\n<\/div>\n

Find out more<\/a><\/p>\n<\/section>\n<\/div>\n

\n
\n
\n

Specification<\/h2>\n
\n

Hailo-10H AI accelerator delivering 40 TOPS (INT4) inferencing performance<\/p>\n

Performance for computer vision models comparable to the Raspberry\u00a0Pi\u00a0AI\u00a0HAT+ (26 TOPS)<\/p>\n

Runs generative AI models efficiently using 8GB on-board RAM<\/p>\n

Fully integrated into Raspberry Pi\u2019s camera software stack<\/p>\n

Conforms to Raspberry Pi HAT+ specification<\/p>\n

Supplied with optional heatsink, and with 16mm stacking header, spacers, and screws to enable fitting on Raspberry Pi 5 with Raspberry\u00a0Pi\u00a0Active\u00a0Cooler in place<\/p>\n<\/div>\n

Documents<\/h2>\n
\n