ASKIX As Your Hardware Provider
For Software Vendors & ISVs
RM-01's Official Canadian Distributor
The hardware layer for the AI product you're shipping.
A portable AI supercomputer for private, on-premises deployment of AI applications.
64GB
LPDDR5 VRAM - local inference, no cloud roundtrip
1000 TOPS
computing power -
2048 CUDA core
80B
max model parameters -
mainstream open-source models
Full Specs
What you're building on
Enterprise-grade AI hardware with complete data sovereignty.
GPU
NVIDIA T234
64GB LPDDR5 VRAM ·
1000 TOPS · 2048 CUDA cores · 204.8 GB/s memory bandwidth
CPU
Intel Core i3-N305
x86 · 8 cores · up to 3.8 GHz
RAM
16GB
LPDDR5 for OS and applications
STORAGE
4TB + 4TB
NVMe M.2 SSD up to 4TB (data) · CFexpress card up to 4TB
(models, applications)
MODELS
Up to 80B
Mainstream open-source
(Gemma, Qwen, Llama, etc.) ·
custom enterprise models
OS
Ubuntu-based
Linux distribution optimized
for AI development
POWER
140W
5A / 28V
PHYSICAL
1.7kg
17 × 4.5 × 25cm
(6.69 × 1.77 × 9.84in) ·
dust & vapour proof​
ENVIRONMENT
−20°C to 60°C
Operating/storage ·
up to 5,000m altitude ·
automotive-grade stability
Integration Path
From your AI application to your customer's site in three steps.
1. Pack your application
Package your AI application as a Docker image - the standard format RM-01 runs.
2. Load the CFexpress card
Load your models and application image onto the CFE card - one card carries everything.
3. Power on and deploy
Insert the card and power on - your app and models stay cryptographically locked to that device.