Pricing

Cloud built for AI teams.

Per-minute billing, no commitments. Boot a Template in 1.8 seconds or a VM in under 90s. Scale to 8 GPUs per instance.

Read the docs
Live capacity across India & EU regions

GPUs

NVIDIA H100, A100 & L4 · per GPU pricing

H100 SXM

Hopper · flagship for 70B+ inference and fine-tuning

$2.69
On-demand · /hr
80 GB VRAM · 200 GB RAM · 16 vCPU

A100 80GB

Ampere · proven for training & inference

$1.49
On-demand · /hr
80 GB VRAM · 112 GB RAM · 16 vCPU

A100 40GB

Ampere · great cost-per-token

$0.89
On-demand · /hr
40 GB VRAM · 112 GB RAM · 16 vCPU

L4

Ada · low-cost inference & notebooks

$0.44
On-demand · /hr
24 GB VRAM · 124 GB RAM · 32 vCPU

On-demand instances are billed per minute. Pause and resume any time. All prices in USD per GPU per hour (₹ shown automatically for India).

CPU VMs

General-purpose compute · on-demand only

2 vCPU · 8 GB RAM

General-purpose VM

$0.05
On-demand · /hr

4 vCPU · 16 GB RAM

General-purpose VM

$0.10
On-demand · /hr

8 vCPU · 32 GB RAM

General-purpose VM

$0.20
On-demand · /hr

16 vCPU · 64 GB RAM

General-purpose VM

$0.40
On-demand · /hr

32 vCPU · 128 GB RAM

General-purpose VM

$0.79
On-demand · /hr

CPU VMs are priced as $0.012 / vCPU + $0.0032 / GB RAM per hour. Available in the India (Noida) region.

Storage$0.10/GB · month
EgressFreein-region
BillingPer minutepause anytime
Need 25+ GPUs?Talk to sales →

Not sure which one?

Templates and VMs share the same GPUs and the same pricing — the difference is what's running on top.

Templates

Managed containers

Pre-built JupyterLab, VS Code, API endpoint, or SSH-only stacks. We manage the OS, drivers, CUDA, and the container — you bring code.

  • Boot in 1.8 seconds
  • Persistent storage across stop / start
  • One-click expose API or notebook URL
  • Up to 8 GPUs per template

Pick this when you want to start training or running notebooks immediately.

On-Demand VMs

Root access & full control

A clean VM with full root and your own kernel or Docker stack. Bring orchestration, custom drivers, daemons, your own CI runners.

  • Root SSH from minute one
  • Bring your own Docker / Kubernetes
  • Custom kernels and drivers
  • Up to 8 GPUs per VM, multi-region

Pick this when you want to full OS control or running infra Templates can't host.

Trusted by 3,000+ researchers and 50+ startups worldwide

99.9% uptime SLAPer-minute billingFree in-region egressIndia + EU regions
Frequently asked questions

Feedback from Our Valued Users

Cagri Demir

I published a paper with the results from exps i did on your gpus when @googlecloud turned me down. Thanks a lot for the great product/service. Not to mention super competitive prices...

Cagri Demir
@adnancagri
Pricing – Cloud built for AI teams | Jarvislabs | Jarvis Labs