OptraCloudAI Inference · powered by Globussoft Request an instance
Per-second billing · SSH in · no reservation

GPUs by the hour, at the cost of running them.

We own the building and the power connection, so our costs start at the electricity bill. There is no colocation invoice underneath the rate. RTX PRO 6000 at $1.39/hr against $1.47 elsewhere in India, billed by the second with a one-minute minimum. SSH key in, CUDA and the whole stack already there.

What an hour actually costs usRTX PRO 6000
Electricity, at Rs9/kWh
$0.06
Card, amortised
$0.66
Same card, rented elsewhere in India
$1.47
Our rate
$1.395% less, same silicon
We own the facility.
Rates

Per second, one-minute minimum, nothing reserved.

Prices exclude GST. Billed in USD or INR — GST invoices on every bill. Stop the instance and billing stops with it.

GPUVRAMBest for Elsewhere in IndiaOur rate
RTX A5000 Ampere · 24 GB GDDR6 ECC 24 GB 7B-14B inference, image gen, dev boxes $0.39per hour
RTX 5090 Blackwell · 32 GB GDDR7 32 GB Fast image gen, 14B-32B inference $0.95 $0.76per hour
RTX PRO 6000 Blackwell 96 GB GDDR7 · FP4 · server edition 96 GB Video generation, 70B inference, fine-tuning $1.47 $1.39per hour
RTX PRO 6000 ×4 384 GB aggregate, one host 384 GB Large MoE, multi-GPU fine-tuning $5.86 $5.39per hour

Comparison rates are the published Indian per-hour prices for the same cards, converted at Rs87. We do not currently offer H100 or H200 by the hour. Provisioning takes under an hour today and a person does it, so the figure quoted here is the one we actually hit.

Pre-installed

SSH in and start. Nothing to build first.

Every image ships with the stack already configured against the driver, so you are not spending the first billed hour compiling CUDA extensions.

CUDA12.8
PyTorch2.7
vLLM0.9
ComfyUIlatest
TensorRT-LLM0.18
Ollama0.7
diffusers0.33
JAX0.5
Transformers4.52
Jupyter Lab4.x
Docker27
conda + git-lfs-

Bring your own image if you prefer — Docker with the NVIDIA runtime is configured, and we will pull a private registry image for you on request. Persistent volumes survive instance restarts and are billed separately at $0.04 per GB per month.

How it works

Three steps, no sales call.

01

Send a key

Tell us the GPU and give us an SSH public key. No credit card to start, no minimum commitment, no reservation.

02

We provision

Under an hour, and you get an IP, a port and a root shell on a machine with the whole stack already on it.

03

Billing stops when you do

Per second, one-minute minimum. Destroy the instance and the meter stops. Invoiced monthly with GST.

Reserve and save

Running it all month? Stop paying by the hour.

An RTX 5090 by the hour is $555 if you run it all month. The dedicated monthly plan is $459 — cheaper than the meter, and the models come loaded and maintained. Past about 600 hours, reserving wins.

If you runBy the hourOn a monthly plan You save
Under 600 hrs a monthfrom $152 stay hourly — it is cheaper
700 hrs a month$532 $459dedicated, models included $73
The whole month, unmetered $555 $459dedicated, models included $96

Raw hourly is cheaper than our Video plan if all you want is the card. The monthly plans cost more because we load and maintain the models, fine-tune on your footage and answer the phone. If you only need silicon, rent it by the hour. See what the monthly plans include.

Who you are buying from

The company behind the hardware.

OptraCloud is the AI infrastructure arm of Globussoft Technologies, incorporated 2012. We run the facility, announce our own address space, and hold the certifications your security review will ask for.

ISO 27001
:2022 certified
ISO 27018
:2019 cloud PII
AS153819
Our own ASN and /23

Certificates are held by Globussoft Technologies Pvt. Ltd. — ISO/IEC 27001:2022 for information security, cyber security and privacy protection (cert IC-58930/1225, valid to December 2028) and ISO/IEC 27018:2019 for personally identifiable information in public cloud (GCS/ITST/8532). We send full certificates and scope statements on request, before you commit to anything. 300+ engineers, serving over two million users across 50 countries since 2012.

Billing

Invoiced the way you need it.

USD OR INR

Your currency, your terms

Billed in dollars or rupees, whichever suits your accounting. Indian customers get GST invoices on every bill with no forex markup and no card-network conversion fee.

NO LOCK-IN

Month to month by default

No reservation, no minimum term, no cancellation fee. Annual prepay earns a discount if you want one, but nothing on this site requires a commitment to get the price you see.

NO SURPRISES

Nothing meters in the background

Flat plans do not meter tokens, render seconds or requests. Hourly billing stops the second you destroy the instance. Overages, where they exist, are published on this page.

Questions we get

Before you ask.

Is my data used for training? No. Nothing you send is logged, retained beyond the request, or used to train anything. On dedicated plans the weights and the machine are yours for the term.
What happens if I outgrow the plan? Move up a tier or add a card. Same endpoint, same API keys, no migration and no re-onboarding.
Can I bring my own model? Yes. Any open-weight model that fits the VRAM, plus your own fine-tunes and LoRAs. Ask and we load it.
Where is the hardware? One facility we own and operate, with dual-carrier transit and our own ASN. If your contract requires a specific jurisdiction, we can commit to it in writing.
What is the SLA? Availability targets go in the contract. Ask during evaluation and we will be specific about what we will sign.
Do you offer a trial? Shared plans start at $199 a month and cancel any time, which is a cheaper way to test quality than any trial we could design. Hourly GPUs start at $0.39.
Get a GPU

Tell us which card and send a key.

We come back with an IP and a port. No sales call unless you want one.

No card required to start. We reply within one business day — [email protected]