GPU Slicing

Rent fractional GPUs (vGPUs).

Access fractional GPU resources. From 1/8 GPU to multi-GPU clusters. NVIDIA A100, H100, and B200. No commitment, pay per second.

H100 80GBUser AUser BUser CUser D

1/8 GPU

Fraction

A100/H100

GPUs

MIG/vGPU

Isolation

Per second

Billing

GPU, sliced.

Fractional GPUs. Per-second billing.

Fractional GPUs

From 1/8 to full GPU. Scale without waste.

A100, H100, B200

Latest NVIDIA GPUs. Up to 80GB HBM3e.

MIG isolation

Hardware-level isolation with NVIDIA MIG technology.

Per-second billing

No minimum commitment. Pay only for what you use.

Auto-scaling

Scale GPUs up and down based on inference load.

Jupyter integration

One-click Jupyter notebooks with GPU attached.

Getting started

Launch your first instance in three steps. CLI, console, or API — your choice.

Terminal
ur gpu launch my-workspace \
  --gpu=a100-40gb --fraction=1/4

GPU patterns.

Model training and inference.

Model fine-tuning

Fine-tune models with fractional GPU for cost savings.

View tutorial

Suggested configuration

1/4 GPU · Per-second · MIG

Estimate your costs

Create detailed configurations to see exactly how much your architecture will cost. Pay for what you use, down to the second.

Configuration 1

Estimated: $44.20/mo

GPU Slice

Usage Volume

hrs

Infrastructure

GB

Options

Premium SLA (99.99%)+25% for guaranteed availability
Config 1 cost$44.20

Cost details

$44.20

Fractional GPU access. No commitment. Pay per hour.

Configuration 1
$44.20
2× standard Replica(s)$29.20
Request Processing$10.00
Storage$5.00

Works seamlessly with

LLM API
Jupyter
K8s
IAM
Monitoring
Analytics

Frequently asked questions

GPU, sliced.

Fractional GPUs. Per-second billing. A100/H100.