Raspberry Pi Launches New AI HAT+ Boards with Up to 26 TOPS Performance

BigGo Editorial Team
Raspberry Pi Launches New AI HAT+ Boards with Up to 26 TOPS Performance

Raspberry Pi continues to strengthen its AI capabilities with the release of two new AI HAT+ boards, offering significant performance improvements and simplified integration for AI applications.

Key Features and Specifications

  • Two Performance Variants:
    • Entry-level model with 13 TOPS (Hailo-8L)
    • High-performance model with 26 TOPS (Hailo-8)
  • Improved Design:
    • Direct PCB integration of Hailo accelerator
    • PCIe Gen 3 interface support
    • Enhanced thermal dissipation
  • Package Contents:
    • 16mm spacers (compatible with Active Cooler)
    • Stacking header
    • Flat flex cable for PCIe connection

Performance and Compatibility

The new AI HAT+ represents a significant evolution from the previous AI Kit design. The 26 TOPS variant particularly stands out with its ability to:

  • Handle larger neural networks
  • Run multiple networks simultaneously
  • Maintain backward compatibility with existing AI Kit software

Software Integration

The boards feature seamless integration with:

  • Latest Raspberry Pi OS
  • Automatic accelerator detection
  • Support for rpicam-apps and Picamera2
  • Note: Models optimized for Hailo-8 may not work on Hailo-8L

Pricing and Availability

  • 13 TOPS model (Hailo-8L): $70
  • 26 TOPS model (Hailo-8): $110
  • Currently available through CanaKit with limited stock
The Raspberry Pi computer, showcasing its compact design and connectivity options, ideal for AI applications highlighted in the new HAT+ board releases
The Raspberry Pi computer, showcasing its compact design and connectivity options, ideal for AI applications highlighted in the new HAT+ board releases

Bottom Line

The new AI HAT+ boards demonstrate Raspberry Pi's commitment to making AI acceleration more accessible and efficient. The simplified design and improved thermal management make these boards particularly attractive for AI developers looking to enhance their Raspberry Pi 5 projects.