← All productsPreview
Private GPUs · Preview · Preview
Compute, Dedicated Lab Inference
Private GPU inference on lab hardware — dedicated nodes for teams that need isolation, predictable latency, or models we host outside shared cloud.
DedicatedPrivateHybridLab nodes
Preview · Einstein lab hardware · not multi-tenant
Compute
Private lab inference
Lab
Einstein GPU node
Dedicated hardware · not multi-tenant
VRAM
82%
Queue
2 jobs
GPU load
74%
Temp
61°C
VRAM · 32 GB26 GB used
GPU load74%
Job queue
geniuspro-mini4 · Ollamarunning
embed-batch · nightlyqueued
vision-ocr · invoicequeued
GPU activitylast 60s
Your hardware · private inference · pairs with Fleet
Capabilities
Isolation when shared cloud isn't enough
Run sensitive workloads on dedicated GPUs, route through Simon when you need hybrid burst, and scope hardware with our team.
Dedicated
Dedicated GPUs
Lab nodes — not multi-tenant shared inference.
Private
Private workloads
For regulated data and custom fine-tunes.
Hybrid
Hybrid with Simon
Route sensitive jobs locally, burst to cloud when needed.
Access
Preview access
Talk to us to scope hardware and SLAs.
Works with