Buy anything from 5,000+ international stores. One checkout price. No surprise fees. Join 2M+ shoppers on Desertcart.
Desertcart purchases this item on your behalf and handles shipping, customs, and support to Hungary.
🚀 Elevate your edge AI game with power, speed, and efficiency!
The Hailo-8 M.2 AI Accelerator Module delivers a powerhouse 26 TOPS AI processing capability with ultra-low 2.5W power consumption. Designed for edge devices like Raspberry Pi 5, it supports major AI frameworks and runs on Linux and Windows. Its rugged design withstands extreme temperatures (-40°C to 85°C), while the PCIe Gen3 interface ensures seamless integration. Ideal for real-time, multi-model AI inferencing in video detection and image analysis, this module transforms your edge computing with unmatched efficiency and speed.
| ASIN | B0FBFTW5QJ |
| Best Sellers Rank | 30,669 in Computers & Accessories ( See Top 100 in Computers & Accessories ) 2,339 in Components & Replacement Parts |
| Box Contents | User Manual |
| Brand Name | Unistorm |
| CPU Model | Others |
| Compatible Devices | Raspberry Pi 5 |
| Customer Reviews | 4.2 4.2 out of 5 stars (11) |
| Manufacturer | Unistorm |
| Manufacturer Part Number | Hailo-8 |
| Model Name | Hailo-8 M.2 AI Accelerator Module |
| Model Number | Hailo-8 |
| Network Connectivity Technology | PCIe Gen3, 4-lane |
| Operating System | Linux, Windows 10 |
| Processor Brand | Hailo |
| Processor Count | 1 |
| RAM Memory Technology | LPDDR4 |
| UPC | 797465078001 |
| Unit Count | 1.0 count |
J**T
I paired it with a SABRENT M.2 NVMe SSD to PCIe to add it to my host with no NVMe slots. It works great with Frigate and takes most of the load off of the CPU. If you are setting this up with a Proxmox server, Debian will see the device and allow you to pass 8t through to the VM, but Proxmox scrubs the PCI device list, so you have to add it's ID manually, but it works
J**.
It works well, and surprisingly fast for inference, with provided code and libraries. Please note that even though the base software is Open Source, some of the code and file formats necessary to use this chip are proprietary.
P**K
Not for LLMs. More for image analysis and inference. Video detection etc.
F**K
Defect in product caused a short circuit that caused a fire. Had this not been setup in my amateur lab with CO2 fire extinguishers, Halon fire extinguishers, sand, and ceramic blankets, the fire could have been a _much_ more serious issue. The odds of most people having the sorts or fire control equipment as I have in my lab are pretty low. Fire in a lab is a very bad thing, and so care is taken to prepare for the possibility. Most people will operate this item under little to no supervision with only basic equipment to respond if it catches on fire.
A**N
So far no problems! Installed into my nvr (Debian hosted) no problem. Figuring out the drivers and documentation took some effort.
Trustpilot
1 month ago
2 months ago