Raspberry Pi launches a board for running generative AI locally

Raspberry Pi is betting on local artificial intelligence with its new expansion board, the AI HAT+ 2. For $130, Raspberry Pi 5 users can now run small generative AI models directly on their device, with no need to rely on cloud computing. It is a significant step forward for enthusiasts and developers who want to experiment with AI offline.

The AI HAT+ 2 is an updated, more powerful version of the earlier AI HAT+. The key difference is 8 GB of RAM and a Hailo 10H chip that delivers up to 40 TOPS (trillion operations per second) of AI performance. The Raspberry Pi 5 can use the AI HAT+ 2 to handle AI-related tasks, leaving the main Arm processor free for other work. The previous version focused mainly on image processing, while the AI HAT+ 2 can run small generative AI models such as Llama 3.2 and DeepSpeech.

Why does this matter? Running AI models locally has several advantages. First, privacy: data never leaves the device, which matters for applications that handle sensitive information. Second, speed: without the delays of sending data to the cloud and back, performance can improve significantly. Third, availability: local AI works even without an internet connection, which is critical for projects in remote or unreliable locations.

The Hailo 10H chip is the key factor in delivering the necessary performance. Hailo specializes in high-performance, energy-efficient AI accelerators for edge computing. Its chips can run complex AI models on resource-constrained devices such as the Raspberry Pi. Integration with the Raspberry Pi opens up new possibilities for AI in robotics, home automation, industrial control and other fields.

What will this mean for the industry and for users? For developers, it means a more accessible platform for experimenting with AI. The Raspberry Pi is already a popular prototyping tool, and the AI HAT+ 2 makes it even more appealing for machine learning projects. For ordinary users, it may lead to new applications and devices that use local AI to improve privacy, speed and availability.

In conclusion, the release of the AI HAT+ 2 is an important step toward democratizing artificial intelligence. By making it possible to run generative AI models locally on a Raspberry Pi, the company opens new horizons for creativity, innovation and practical use of machine learning. It may become a catalyst for new and exciting AI applications across a wide range of fields.