Specifications
CPU
Intel 14th / 13th / 12th Gen Core processors, LGA1700 socket, max 35W TDP (default i5-14500T)
AI Accelerator
ARES-1983H-AI-MX: 1 x MemryX MX3 M.2 (24 TFLOPS); ARES-1983H-AI-2MX: 2 x MemryX MX3 (48 TFLOPS); ARES-1983H-AI-4MX: 4 x MemryX MX3 (96 TFLOPS)
Memory
1 x 262-pin DDR5 SO-DIMM, 4800/5600MHz, up to 32GB (8GB default on MX/2MX, 16GB on 4MX)
Chipset
Intel H610E (MX / 2MX), Intel Q670E (4MX)
Graphics
Integrated Intel UHD Graphics 730 / 770
Storage
2.5in HDD/SSD bay with 256GB default (MX); M.2 M-Key 2242/2280 PCIe 4.0 x4 + SATA with 256GB default (2MX / 4MX)
LAN
1 x 2.5GbE port plus 2 x GbE ports
LAN Chipset
1 x Intel i226LM, 1 x Intel i219LM, 1 x Intel i210AT
USB
2 x USB-A 3.2 Gen2 10Gbps (MX/2MX) or 4 x USB-A 3.2 Gen2 (4MX), 2 x USB-A 3.2 Gen1, 2 x USB-A 2.0, 2 x internal USB-A 2.0
Serial Ports
1 x RS232, 1 x RS232 (default) / 422 / 485
Video Output
1 x DisplayPort 1.4 (DP++) up to 4096x2304@60Hz, 1 x VGA up to 1920x1200@60Hz
Expansion
1 x M.2 E-Key 2230 (PCIe 3.0 x1 + USB 2.0 + CNVi), 1 x mPCIe full-size (PCIe 3.0 x1 + USB 2.0) with nano SIM
TPM / Watchdog
dTPM 2.0; watchdog 1~255 levels reset
Power Input
DC 12 ~ 24V via 2-pin terminal block
Power Consumption
95W (MX), 110W (2MX), 140W (4MX) maximum
Operating Temperature
-5 ~ 50°C ambient with airflow; -20 ~ 50°C with industrial RAM and SSD
Storage Temperature
-30 ~ 70°C
Humidity
10 ~ 95% @ 60°C non-condensing
Shock & Vibration
IEC 68-2-64 vibration 3 Grms 5-500Hz; IEC 68-2-27 shock 10G operating / 30G non-operating
Certification
CE, FCC Class A
Construction / Mounting
Metal and aluminium, wall-mount standard, DIN-rail optional with CTOS bracket
Dimensions
85 x 210 x 164mm (MX); 95 x 210 x 164mm (2MX / 4MX)
Weight
3.6 kg (MX), 3.9 kg (2MX), 4.3 kg (4MX)
OS Support
Windows 10 / Windows 11 / Ubuntu 22.04 / Ubuntu 24.04
Overview
The ARBOR ARES-1983H-AI Series is a compact industrial edge AI computer that scales inference throughput by the number of installed accelerators rather than by changing CPU class. A single LGA1700 socket accepts 12th, 13th or 14th Generation Intel Core processors at up to 35W TDP, and up to four MemryX MX3 M.2 modules add 24 TFLOPS each for a 96 TFLOPS ceiling.
Three SKUs ship pre-configured: ARES-1983H-AI-MX with one accelerator and Intel H610E, ARES-1983H-AI-2MX with two accelerators, and ARES-1983H-AI-4MX with four accelerators on Intel Q670E. All three take DDR5 SO-DIMM memory up to 32GB, offer 2.5GbE plus dual GbE networking, and provide RS-232/422/485 serial, multiple USB 3.2 Gen2 ports and DisplayPort plus VGA output.
The MX3 accelerators are dedicated inference engines: they add TOPS without the thermal envelope or power draw of a discrete GPU, which keeps the system within a 95-140W total budget and allows a compact 85-95mm wide wall-mount or DIN-rail enclosure.
This makes the series a good fit for multi-camera vision inspection, defect detection and industrial video analytics where predictable power, deterministic thermal behaviour and long-life Intel platform support are required. QS Compute supplies all variants with industrial-grade memory and storage options.
Key Benefits
Scales from 24 to 96 TFLOPS by adding MemryX MX3 M.2 accelerators; Intel 14th/13th/12th Gen Core CPU performance in the same chassis; total system power of only 95-140W with no discrete GPU; 2.5GbE plus dual GbE for camera feeds; industrial -20 to 50C rating with industrial memory and storage; wall-mount or optional DIN-rail installation; Ubuntu 22.04/24.04 and Windows 10/11 support.
Applications
Multi-camera machine vision inspection; defect and anomaly detection; industrial video analytics; smart manufacturing quality control; automated optical inspection (AOI); edge AI inference for medical and logistics imaging.
Request a Quote — ARBOR ARES-1983H-AI SERIES — COMPACT EDGE AI COMPUTER WITH MEMRYX MX3 UP TO 96 TFLOPS
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →