Component guide · Intermediate · 3 min read
Raspberry Pi AI HAT+

13 TOPS Hailo-8L, 26 TOPS Hailo-8 and AI HAT+ 2 with Hailo-10H for Raspberry Pi 5
By Geekobot Learning Team · Updated 29 Oct 2026
On this page
The AI HAT+ adds a Hailo neural network accelerator to the Raspberry Pi 5 through its PCIe port. The 13 TOPS Hailo-8L runs real-time object detection with a Pi camera; the 26 TOPS Hailo-8 handles bigger models or several streams; and the AI HAT+ 2 with Hailo-10H adds its own memory so it can run small language and vision-language models as well.
Key specs at a glance
- Accelerator
- Hailo-8L (13 TOPS), Hailo-8 (26 TOPS), Hailo-10H (40 TOPS INT4)
- Interface
- PCIe Gen 3 x1 via FFC
- Memory
- 8 GB on AI HAT+ 2
- Board
- Raspberry Pi 5 only
- Software
- hailo-all package, rpicam-apps, Picamera2
- Power
- Few watts; use the 27W supply
Models in this guide
These are alternatives. Pick the one that fits your project; you only need one.
Option 1Raspberry Pi AI HAT+ with Hailo 8L (13 TOPS)₹7,839
Option 2Raspberry Pi AI HAT+ 26 TOPS₹12,339
Option 3Raspberry Pi AI HAT+ 2 – Hailo-10H AI Accelerator for Pi 5₹26,709
What the Raspberry Pi AI HAT+ (Hailo-8L, Hailo-8 and Hailo-10H) is
The Raspberry Pi AI HAT+ and AI Kit use a Hailo-8L (13 TOPS) or Hailo-8 (26 TOPS) accelerator on the Pi 5's PCIe port to run object detection, pose and segmentation in real time.
Install with sudo apt install hailo-all (or the h10 package for the AI HAT+ 2) and the camera demos run detection, pose and segmentation at 30 fps. Models must be compiled to Hailo's HEF format; many YOLO models come ready-made.
It does not speed up normal Python code; only models compiled for Hailo run on it.
Raspberry Pi AI HAT+ (Hailo-8L, Hailo-8 and Hailo-10H) pinout
The pins on the Raspberry Pi AI HAT+ (Hailo-8L, Hailo-8 and Hailo-10H) and what each one does.
| Pin | Function |
|---|---|
| PCIe FFC | Pi 5 PCIe connector |
| GPIO header | Passes through; HAT+ EEPROM |
| Heatsink | On the Hailo chip |
| Standoffs | Fit over the active cooler |
- PCIe FFC
- Pi 5 PCIe connector
- GPIO header
- Passes through; HAT+ EEPROM
- Heatsink
- On the Hailo chip
- Standoffs
- Fit over the active cooler
Wiring the Raspberry Pi AI HAT+ (Hailo-8L, Hailo-8 and Hailo-10H)
Fit the active cooler first, then the stacking header and standoffs, then the HAT. Enable PCIe Gen 3 for full speed.
| Raspberry Pi AI HAT+ (Hailo-8L, Hailo-8 and Hailo-10H) pin | Connection | Note |
|---|---|---|
| FFC | Pi 5 PCIe port to HAT | Contacts toward pins |
| Header | Press onto GPIO with spacer | Active cooler underneath |
| Camera | CAM port | 22-pin cable |
- FFC
- Connection:
- Pi 5 PCIe port to HAT
- Note:
- Contacts toward pins
- Header
- Connection:
- Press onto GPIO with spacer
- Note:
- Active cooler underneath
- Camera
- Connection:
- CAM port
- Note:
- 22-pin cable
Example code
Raspberry Pi OS sets PCIe Gen 3 automatically for the AI HAT+. The hailo-rpi5-examples repository has Python pipelines for detection, pose and segmentation.
sudo apt update && sudo apt full-upgrade -y
sudo apt install hailo-all # AI HAT+ (Hailo-8L / Hailo-8)
sudo reboot
hailortcli fw-control identify # shows the Hailo chip
rpicam-hello -t 0 --post-process-file \
/usr/share/rpi-camera-assets/hailo_yolov8_inference.jsonSetup and tuning
connect the HAT to the Pi 5 PCIe port with the FPC cable and stand-offs.
install the hailo-all package and enable PCIe Gen 3 for full speed.
run the rpicam-apps Hailo demos or the hailo-rpi5-examples repository.
use models compiled to HEF format for Hailo; convert others with the Hailo Dataflow Compiler.
fit the active cooler on the Pi 5; the accelerator also warms.
What you need
Shop this list7 of 7 found in our store · ₹51,740
Raspberry Pi AI HAT+ with Hailo 8L (13 TOPS): the part this guide coversRaspberry Pi AI HAT+ with Hailo 8L (13 TOPS)₹7,839
Raspberry Pi AI HAT+ 26 TOPS: another version this guide coversRaspberry Pi AI HAT+ 26 TOPS₹12,339
Raspberry Pi AI HAT+ 2 – Hailo-10H AI Accelerator for Pi 5: another version this guide coversRaspberry Pi AI HAT+ 2 – Hailo-10H AI Accelerator for Pi 5₹26,709
Raspberry Pi 5 (4 GB): the board the HAT stacks onGeekworm Raspberry Pi 5 PoE HAT 802.3af/at (USB-C)₹2,659Closest match, check the specs
Official 27W USB-C power supply for Pi 5: steady power for the PiOfficial 27W USB-C PD Power Supply for Raspberry Pi 5 – IN Plug₹1,319
Official Raspberry Pi 5 Active Cooler: needed under AI workloadsOfficial Raspberry Pi 5 Active Cooler₹566
15-pin to 22-pin camera cable: to add a cameraArducam B0087 15-Pin to 22-Pin Camera Cable for Pi Zero v1.3 (Pack of 2)₹309
Alternatives and when to choose them
- AI vision sensors: no programming needed.
- Depth cameras: OAK has its own AI chip.
- Dual NVMe and PCIe boards: AI plus SSD.
Common mistakes to avoid
- Leaving PCIe at Gen 2.
- Trying to run uncompiled models.
- Running without the active cooler.
- Expecting it to work on a Pi 4.
Frequently asked questions
What can the Raspberry Pi AI HAT do?
It runs YOLO detection, pose estimation and segmentation at real-time frame rates.
Does it work on a Pi 4?
No. It needs the PCIe port that only the Pi 5 has.
Which version should I buy, 13 or 26 TOPS?
26 TOPS runs larger models or more streams; 13 TOPS is enough for most single cameras.
Can I run my own model?
Yes, after compiling it to HEF with Hailo's tools.
Can I use the AI HAT and an NVMe SSD together?
Only with a PCIe switch board made for both.
Parts for this guide
Keep learning
Sources
Written by the Geekobot team from manufacturer datasheets. Always check the datasheet for the exact part and variant you buy.


