---
title: Rent cloud GPUs by the second, H100 from $1.85/h | GPUBank
url: https://gpubank.io/
description: Rent H100, H200, B200, A100 and RTX 4090 GPUs by the second in Paris, Frankfurt, Ashburn and Singapore. CPU, RAM, NVMe and traffic included. Pay in crypto.
generated: 2026-11-05T16:50Z
publisher: GPUBank (GPU Compute OÜ, Tallinn, Estonia)
summary_for_ai: https://gpubank.io/llms.txt
full_text_for_ai: https://gpubank.io/llms-full.txt
---

> Markdown version of https://gpubank.io/, generated from the live page. GPUBank is a GPU cloud billed by the second at one all-in price per GPU (vCPU, RAM, NVMe, a public IPv4 and traffic included), in Paris, Frankfurt, Ashburn and Singapore. Prices are in US dollars, excluding VAT.

Cloud GPU rental, billed per second

# Rent cloud GPUs by the second, at the price you saw.

Rent an H100 SXM for $1.85 an hour, with the vCPUs, RAM, NVMe storage, a public IPv4 address and all your traffic already in the price. Billing starts once the machine passes its health check and ends the second you stop it.

Launch a GPU

See live availability

Top up from $25 in crypto and pay only for the seconds you use, with no minimum and no commitment.

**Allocatable now**

1,513 GPUs

updated 4 s ago

**Median launch**

48 s

p95 71 s · last 30 days

**H100 SXM**

$1.85 /h

$0.000514 / s · all-in

**Regions**

Paris · Frankfurt · Ashburn · Singapore

In our racks today

## 10 GPU models in 4 regions.

Every card has one all-in price, the same in each region, billed by the second. The count under each price is what you can launch right now, all regions together.

See live availability

- NVIDIA B200 SXM 180 GB $4.15 /h **−47%** cheaper avg market price **$7.94** 35 available
- NVIDIA H200 SXM 141 GB $2.49 /h **−44%** cheaper avg market price **$4.51** 69 available
- NVIDIA H100 SXM 80 GB $1.85 /h **−49%** cheaper avg market price **$3.67** 316 available
- NVIDIA H100 PCIe 80 GB $1.35 /h **−50%** cheaper avg market price **$2.71** 87 available
- NVIDIA RTX PRO 6000 96 GB $1.15 /h **−49%** cheaper avg market price **$2.27** 132 available
- NVIDIA A100 SXM 80 GB $0.95 /h **−50%** cheaper avg market price **$1.93** 229 available
- NVIDIA L40S 48 GB $0.55 /h **−55%** cheaper avg market price **$1.24** 280 available
- AMD MI300X 192 GB $1.65 /h **−57%** cheaper avg market price **$3.91** 39 available
- NVIDIA RTX 5090 32 GB $0.48 /h **−14%** cheaper avg market price **$0.56** 52 available
- NVIDIA RTX 4090 24 GB $0.23 /h **−45%** cheaper avg market price **$0.42** 274 available

45 % Cheaper than the average GPU cloud, on the same card.

Forget the overhyped clouds that bill you for their brand. Same silicon, same hour, without the markup.

Our all-in hourly price against the public on-demand prices of up to 43 providers for each model, checked on 29 Sep 2026. Spot and reserved offers excluded.

Market average

**$100**

GPUBank

**$55**

[Compare us to 63 GPU clouds](https://gpubank.io/compare)

01Ready-to-run AI templates

## Start from a template and have your stack running in two minutes.

PyTorch, vLLM, ComfyUI, Ollama and twenty other images come pinned to a tested driver and CUDA version, and we launch every one of them on real GPUs every three days to make sure they still work. Pick one and the launcher opens with everything already filled in.

[All 25 templates](https://console.gpubank.io/templates)

### PyTorch 2.8 · CUDA 12.8

pytorch:2.8-cu128

PyTorch with CUDA, cuDNN and NCCL pinned to versions that work together, plus Jupyter and SSH ready to use.

verified Nov 4

2m 01s

8888 · 22

8 GB+

[Launch](https://console.gpubank.io/launch?template=pytorch%3A2.8-cu128)

### Ubuntu 24.04 · CUDA 12.8

ubuntu:24.04-cu128

A clean Ubuntu with the NVIDIA driver, the CUDA runtime and SSH, for when you would rather install your own stack.

verified Nov 4

48s

22

[Launch](https://console.gpubank.io/launch?template=ubuntu%3A24.04-cu128)

### JupyterLab 4

jupyter:4.3-cu128

JupyterLab with a terminal, and the GPU already visible to PyTorch and JAX from the first cell.

verified Nov 5

1m 04s

8888 · 22

4 GB+

[Launch](https://console.gpubank.io/launch?template=jupyter%3A4.3-cu128)

### ComfyUI · model sync

comfyui:0.3-cu128

ComfyUI for Stable Diffusion and FLUX with popular custom nodes installed and your model folder synced from your disk.

verified Nov 5

~4 min

8188 · 22

12 GB+

[Launch](https://console.gpubank.io/launch?template=comfyui%3A0.3-cu128)

### vLLM 0.11 serving

vllm:0.11-cu128

Serve an LLM behind an OpenAI-compatible API, with weights loaded from our regional cache or from Hugging Face.

verified Nov 3

2m 31s

8000

16 GB+

[Launch](https://console.gpubank.io/launch?template=vllm%3A0.11-cu128)

### SGLang

sglang:0.4-cu128

Fast structured generation behind an OpenAI-compatible API, with RadixAttention caching for repeated prompts.

verified Nov 4

2m 44s

30000

16 GB+

[Launch](https://console.gpubank.io/launch?template=sglang%3A0.4-cu128)

### Ollama

ollama:0.12

Pull an open model such as Llama, Qwen or Mistral and chat with it over the API, with models kept on your disk.

verified Nov 3

~2 min

11434

8 GB+

[Launch](https://console.gpubank.io/launch?template=ollama%3A0.12)

### Axolotl fine-tuning Job

axolotl:0.9-cu128

Fine-tune an LLM with LoRA, QLoRA or a full fine-tune from a single YAML file, with checkpoints written to your disk.

verified Nov 3

2m 04s

22

24 GB+

[Launch](https://console.gpubank.io/launch?template=axolotl%3A0.9-cu128)

### Unsloth

unsloth:2025.9-cu128

LoRA fine-tuning about twice as fast on models from 8B to 70B parameters, driven from a notebook.

verified Nov 5

2m 00s

8888 · 22

16 GB+

[Launch](https://console.gpubank.io/launch?template=unsloth%3A2025.9-cu128)

### faster-whisper API

whisper:faster-1.2

Speech-to-text with Whisper large-v3 in int8, batched behind an HTTP API that returns word timestamps.

verified Nov 4

~1 min

9000

6 GB+

[Launch](https://console.gpubank.io/launch?template=whisper%3Afaster-1.2)

### ComfyUI · video (Wan 2.2)

comfyui-video:0.3-wan2.2

Text-to-video and image-to-video workflows built on Wan 2.2, with the weights loaded from our regional cache.

verified Nov 4

4m 12s

8188 · 22

24 GB+

[Launch](https://console.gpubank.io/launch?template=comfyui-video%3A0.3-wan2.2)

[Your Docker image Run any Docker image from a public or private registry, with the GPU runtime added for you. Bring your image](https://console.gpubank.io/launch?template=custom)

02Four steps, about five minutes

## From sign-up to a running GPU in four steps.

No sales call and no contract: sign up, add credit, pick a card, and you are connected about five minutes later.

1. ### Step 1: Create an account
  An email and a password, or GitHub or Google.
  30 s
  you@company.com **confirmed**
2. ### Step 2: Top up your balance
  Bitcoin, USDT, Ethereum, Monero or 12 other coins.
  ~3 min
  $100 paid **$135 credited**
3. ### Step 3: Pick a GPU and a template
  10 GPU models in 4 regions, 25 ready images.
  1 min
  H100 SXM Paris PyTorch
4. ### Step 4: Connect and run
  SSH, Jupyter or HTTPS, billed by the second.
  48 s
  **Running** 00:42 $0.0216

- **5 min** from sign-up to SSH
- Billing starts after the health check
- Stop anytime, your disk stays

[Create your account](https://console.gpubank.io/register)

[Read the quickstart](https://gpubank.io/docs#quickstart)

03Live GPU availability and prices

## Every GPU on this board can be launched right now.

Stock is counted per region and refreshed every 30 seconds. When a region shows queue, that GPU is sold out there for now, and you learn it here instead of after a failed launch.

| GPU | Memory | vCPU | RAM | NVMe | Paris | Frankfurt | Ashburn | Singapore | Price per hour · per second | Launch |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| B200 SXMnew Blackwell | 180 GB | 28 | 283 GB | 2 TB | — | 12 | 23 | — | $4.15 $0.001153 / s | [Launch](https://console.gpubank.io/launch?gpu=b200-sxm) |
| Memory bandwidth 8 TB/s 180 GB HBM3e NVLink 1.8 TB/s GPU to GPU inside the node Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| H200 SXM Hopper | 141 GB | 24 | 276 GB | 2 TB | 37 | — | 19 | 13 | $2.49 $0.000692 / s | [Launch](https://console.gpubank.io/launch?gpu=h200-sxm) |
| Memory bandwidth 4.8 TB/s 141 GB HBM3e NVLink 900 GB/s GPU to GPU inside the node Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| H100 SXM Hopper | 80 GB | 20 | 125 GB | 2 TB | 88 | 60 | 138 | 30 | $1.85 $0.000514 / s | [Launch](https://console.gpubank.io/launch?gpu=h100-sxm) |
| Memory bandwidth 3.35 TB/s 80 GB HBM3 NVLink 900 GB/s GPU to GPU inside the node Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| H100 PCIe Hopper | 80 GB | 16 | 188 GB | 1 TB | — | 39 | 48 | — | $1.35 $0.000375 / s | [Launch](https://console.gpubank.io/launch?gpu=h100-pcie) |
| Memory bandwidth 2.0 TB/s 80 GB HBM2e Interconnect PCIe Gen5 PCIe between cards, up to 8 per machine Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| RTX PRO 6000 Blackwell | 96 GB | 16 | 188 GB | 2 TB | 71 | — | 54 | 7 left | $1.15 $0.000319 / s | [Launch](https://console.gpubank.io/launch?gpu=rtx-pro-6000) |
| Memory bandwidth 1.8 TB/s 96 GB GDDR7 ECC Interconnect PCIe Gen5 PCIe between cards, up to 8 per machine Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| A100 SXM Ampere | 80 GB | 16 | 125 GB | 1 TB | 46 | 75 | 89 | 19 | $0.95 $0.000264 / s | [Launch](https://console.gpubank.io/launch?gpu=a100-sxm) |
| Memory bandwidth 2.0 TB/s 80 GB HBM2e NVLink 600 GB/s GPU to GPU inside the node Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| L40S Ada Lovelace | 48 GB | 16 | 94 GB | 1 TB | 111 | 61 | 69 | 39 | $0.55 $0.000153 / s | [Launch](https://console.gpubank.io/launch?gpu=l40s) |
| Memory bandwidth 864 GB/s 48 GB GDDR6 ECC Interconnect PCIe Gen4 PCIe between cards, up to 8 per machine Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| MI300X AMD CDNA 3 | 192 GB | 24 | 240 GB | 2 TB | — | — | 39 | — | $1.65 $0.000458 / s | [Launch](https://console.gpubank.io/launch?gpu=mi300x) |
| Memory bandwidth 5.3 TB/s 192 GB HBM3 Interconnect Infinity Fabric AMD GPU to GPU link Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| RTX 5090 Blackwell | 32 GB | 9 | 35 GB | 500 GB | 33 | — |  | 19 | $0.48 $0.000133 / s | [Launch](https://console.gpubank.io/launch?gpu=rtx-5090) |
| Memory bandwidth 1.8 TB/s 32 GB GDDR7 Interconnect no NVLink PCIe between cards, up to 4 per machine Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |
| RTX 4090 Ada Lovelace | 24 GB | 8 | 41 GB | 500 GB |  | 106 | 168 | — | $0.23 $0.000064 / s | [Launch](https://console.gpubank.io/launch?gpu=rtx-4090) |
| Memory bandwidth 1.0 TB/s 24 GB GDDR6X Interconnect no NVLink PCIe between cards, up to 4 per machine Networking Public IPv4 included All traffic in and out, unmetered Persistent disk $0.03 per GB-month Keeps your data between runs |  |  |  |  |  |  |  |  |  |  |

B200 SXMnew

Blackwell · 180 GB HBM3e

$4.15 /h

28 vCPU 283 GB RAM 2 TB NVMe 8 TB/s

- Paris —
- Frankfurt 12
- Ashburn 23
- Singapore —

[Launch B200 SXM](https://console.gpubank.io/launch?gpu=b200-sxm)

H200 SXM

Hopper · 141 GB HBM3e

$2.49 /h

24 vCPU 276 GB RAM 2 TB NVMe 4.8 TB/s

- Paris 37
- Frankfurt —
- Ashburn 19
- Singapore 13

[Launch H200 SXM](https://console.gpubank.io/launch?gpu=h200-sxm)

H100 SXM

Hopper · 80 GB HBM3

$1.85 /h

20 vCPU 125 GB RAM 2 TB NVMe 3.35 TB/s

- Paris 88
- Frankfurt 60
- Ashburn 138
- Singapore 30

[Launch H100 SXM](https://console.gpubank.io/launch?gpu=h100-sxm)

H100 PCIe

Hopper · 80 GB HBM2e

$1.35 /h

16 vCPU 188 GB RAM 1 TB NVMe 2.0 TB/s

- Paris —
- Frankfurt 39
- Ashburn 48
- Singapore —

[Launch H100 PCIe](https://console.gpubank.io/launch?gpu=h100-pcie)

RTX PRO 6000

Blackwell · 96 GB GDDR7 ECC

$1.15 /h

16 vCPU 188 GB RAM 2 TB NVMe 1.8 TB/s

- Paris 71
- Frankfurt —
- Ashburn 54
- Singapore 7 left

[Launch RTX PRO 6000](https://console.gpubank.io/launch?gpu=rtx-pro-6000)

A100 SXM

Ampere · 80 GB HBM2e

$0.95 /h

16 vCPU 125 GB RAM 1 TB NVMe 2.0 TB/s

- Paris 46
- Frankfurt 75
- Ashburn 89
- Singapore 19

[Launch A100 SXM](https://console.gpubank.io/launch?gpu=a100-sxm)

L40S

Ada Lovelace · 48 GB GDDR6 ECC

$0.55 /h

16 vCPU 94 GB RAM 1 TB NVMe 864 GB/s

- Paris 111
- Frankfurt 61
- Ashburn 69
- Singapore 39

[Launch L40S](https://console.gpubank.io/launch?gpu=l40s)

MI300X

AMD CDNA 3 · 192 GB HBM3

$1.65 /h

24 vCPU 240 GB RAM 2 TB NVMe 5.3 TB/s

- Paris —
- Frankfurt —
- Ashburn 39
- Singapore —

[Launch MI300X](https://console.gpubank.io/launch?gpu=mi300x)

RTX 5090

Blackwell · 32 GB GDDR7

$0.48 /h

9 vCPU 35 GB RAM 500 GB NVMe 1.8 TB/s

- Paris 33
- Frankfurt —
- Ashburn
- Singapore 19

[Launch RTX 5090](https://console.gpubank.io/launch?gpu=rtx-5090)

RTX 4090

Ada Lovelace · 24 GB GDDR6X

$0.23 /h

8 vCPU 41 GB RAM 500 GB NVMe 1.0 TB/s

- Paris
- Frankfurt 106
- Ashburn 168
- Singapore —

[Launch RTX 4090](https://console.gpubank.io/launch?gpu=rtx-4090)

Available 10 or fewer left Queue, sold out here —Not offered in this region

[Machine-readable price feed](https://gpubank.io/prices.json)

Prices in US dollars, excluding VAT. The price includes the vCPUs, RAM, NVMe, one public IPv4 address and all traffic; a persistent disk at $0.03 per GB per month is the only possible extra.

04Open-model benchmarks

## 412 open models, sized and priced on every GPU we rent.

Chat, code, vision, image, video, speech, voice, music, embeddings and 3D, from 80 labs. Pick a model to see which cards it fits on, how fast it runs and what its output costs, then launch it with the command shown.

**Open models**

412

**Workloads**

12

**GPUs**

10

**Setups sized**

3,844

Browse by lab 80 labs

412 models

· released · likes on Hugging Face

**Parameters**

**Context**

**Weights**

**License**

Best value Fastest Cards needed and precision under each GPU

Best value Fastest

Model card

```

```

**How these figures are made.** They are estimates. Speed and cost come from each card's memory bandwidth and compute and each model's architecture (parameters, active experts, attention and cache), with efficiency factors calibrated on published vLLM, SGLang and diffusers results. Language models: a 2,000-token prompt and a 500-token answer, up to 64 requests at once, with prompt processing counted in the cost. Images: 1024 × 1024 at the model's usual step count. Costs use today's on-demand prices and leave out your disk. Measured runs will replace these figures as we publish them.

[All figures as JSON](https://gpubank.io/benchmarks.json) Updated 4 Oct 2026

05Per-second GPU billing

## You pay for the seconds your GPU runs, and your statement shows every one of them.

Billing starts when your instance passes its health check and stops the second you shut it down. The launch is free, a stopped instance only costs its disk, and every line of your usage has a date, a duration and an amount.

H100 SXM · eu-paris · inst_7c1e

stopped, disk deleted 09-30

$4.13 · 2 h 14 m 03 s · $0.000514 / s

- Not billed launch and health check
- Billed per second GPU running
- Disk only $0.03 per GB-month
- Nothing once deleted

### Usage for September 2026

[Export CSV](https://console.gpubank.io/billing/statement)

Usage API

| Date | Resource | Region | Duration | Rate | Amount |
| --- | --- | --- | --- | --- | --- |
| 09-03 | H100 SXM | eu-paris | 2 h 14 m 03 s | $1.85 / h | $4.13 |
| 09-03 | Disk 200 GB | eu-paris | 27 days | $0.03 / GB-month | $5.33 |
| 09-08 | RTX 4090 | eu-frankfurt | 41 m 12 s | $0.23 / h | $0.16 |
| 09-11 | A100 SXM | us-ashburn | 6 h 00 m 00 s | $0.95 / h | $5.70 |
| 09-11 | Credit, host failure | us-ashburn | session in full | — | −$5.70 |
| 09-15 | L40S | ap-singapore | 3 h 30 m 00 s | $0.55 / h | $1.93 |
| Total September | $11.55 |  |  |  |  |

GET /v1/usage?instance=inst_7c1e

```
{
  "instance": "inst_7c1e",
  "gpu": "h100-sxm",
  "region": "eu-paris",
  "launched_at": "2026-09-03T14:01:23Z",
  "billed_from": "2026-09-03T14:02:11Z",
  "billed_to": "2026-09-03T16:16:14Z",
  "billed_seconds": 8043,
  "usd_per_second": 0.000514,
  "amount_usd": 4.13
}
```

The launch at 14:01:23 and the health check are in the log, not in the bill.

06What happens when something fails

## A server can fail, but your data and your money stay safe.

Hardware breaks in every datacenter, so we decided in advance what happens when it does and what you pay. Your persistent disk lives apart from the host, so instances can move while your data stays where it is.

31 s

**interruption**

0 bytes

**of data lost**

$3.70

**credited, the whole session**

1. 11:42:07
  GPU fault detected on host par1-g03 (Xid 79)
  The host stopped responding to the GPU health probe and was marked unhealthy.
2. 11:42:08
  instance inst_91b3 paused, billing stopped
  The meter stops the same second, not at the end of a billing period.
3. 11:42:25
  replacement host par1-g12 allocated in eu-paris
  A healthy card of the same model, in the same region.
4. 11:42:34
  disk vol-2207 (200 GB) reattached, intact
  The persistent disk was never on the failed host, only attached to it.
5. 11:42:38
  instance running again, billing resumed
  Same disk, same image, same public IPv4 address.
6. 11:43:00
  session credited in full: $3.70 for 2 h 00 m
  A host failure refunds the whole interrupted session, automatically.
7. 11:45:00
  incident posted on the status page
  With a post-mortem once the card has been inspected.

[How interruptions and credits work](https://gpubank.io/terms#availability)

View the status page

07Launch from the console, the CLI or the API

## One command gives you a running GPU in under a minute, and you pay nothing until it is ready.

bash

$

✓

allocated GPU H100 SXM in eu-paris

12.34 s

✓

attached disk mydisk (200 GB)

28.11 s

✓

health check passed, CUDA 12.8

45.02 s

✓

ready, billing started at $0.000514 / s

47.80 s

connect:

ssh root@par1-g07.gpubank.io

H100 SXM · eu-paris

$0.000514 / s

Start from the web console, your terminal, a Python script or Terraform. We allocate the card, attach your disk and run the health check, and the meter only starts once you can connect.

From the command to SSH billing starts at 47.8 s

Not billed

Billed

1. 0 s Command
2. GPU allocated, 12.34 s
3. Disk attached, 28.11 s
4. Health check passed, 45.02 s
5. 47.80 s Ready to connect

Allocation, disk and health check run before the meter. A launch that fails its check costs nothing.

Median launch, last 30 days

48 s · p95 71 s

Run it now

[Get your API key](https://console.gpubank.io/register)

Free account, first top-up from $25.
Nothing is billed before the health check.

- SSH
  with your own keys, on a public IPv4
- JupyterLab and VS Code
  in the browser, or Remote over SSH
- CLI and Python SDK
  script launches, stops and disk moves
- REST API and webhooks
  ready, interruption and budget events
- Terraform provider
  instances, disks and spend caps as code
- Private registries
  and S3-compatible storage for datasets

08GPU cloud in Europe, the US and Asia

## Four regions, and your data follows you from one to the next.

Your compute, disks and backups stay in the region you choose, which keeps data in the EU when it has to stay there. When you need GPUs elsewhere, your disk moves across at a measured speed and you launch with the same keys, image and prices.

Paris eu-paris

Frankfurt eu-frankfurt

Ashburn us-ashburn

Singapore ap-singapore

### Latency between regions

Round trip in milliseconds, measured every hour.

| From / to | Paris PAR eu-paris | Frankfurt FRA eu-frankfurt | Ashburn ASH us-ashburn | Singapore SIN ap-singapore |
| --- | --- | --- | --- | --- |
| Paris PAR | — | 9 ms | 78 ms | 162 ms |
| Frankfurt FRA | 9 ms | — | 86 ms | 158 ms |
| Ashburn ASH | 78 ms | 86 ms | — | 218 ms |
| Singapore SIN | 162 ms | 158 ms | 218 ms | — |

Which region

- **Users in Europe**Paris or Frankfurt, 9 ms apart. EU data stays in the EU.
- **Users in the Americas**Ashburn, 78 ms from Paris, with every one of the ten GPU models.
- **Changing your mind later**A disk moves to another region with one command, at 1.6 GB/s to Frankfurt.

### Paris

eu-paris

EU West

**Allocatable now**

386 GPUs

**Network**

100 Gbps per node

**Disk move**

1.6 GB/s toFrankfurt

**Residency**

EU data stays in the EU

H200 SXM

H100 SXM

RTX PRO 6000

A100 SXM

L40S

RTX 5090

RTX 4090

[Launch in Paris from $0.23/h](https://console.gpubank.io/launch?region=eu-paris)

### Frankfurt

eu-frankfurt

EU Central

**Allocatable now**

353 GPUs

**Network**

100 Gbps per node

**Disk move**

1.6 GB/s toParis

**Residency**

EU data stays in the EU

B200 SXM

H100 SXM

H100 PCIe

A100 SXM

L40S

RTX 4090

[Launch in Frankfurt from $0.23/h](https://console.gpubank.io/launch?region=eu-frankfurt)

### Ashburn

us-ashburn

US East

**Allocatable now**

647 GPUs

**Network**

100 Gbps per node

**Disk move**

0.9 GB/s toParis

**Residency**

US data stays in the US

B200 SXM

H200 SXM

H100 SXM

H100 PCIe

RTX PRO 6000

A100 SXM

L40S

MI300X

RTX 5090

RTX 4090

[Launch in Ashburn from $0.23/h](https://console.gpubank.io/launch?region=us-ashburn)

### Singapore

ap-singapore

APAC

**Allocatable now**

127 GPUs

**Network**

100 Gbps per node

**Disk move**

0.4 GB/s toParis

**Residency**

Data stays in Singapore

H200 SXM

H100 SXM

A100 SXM

L40S

RTX 5090

[Launch in Singapore from $0.48/h](https://console.gpubank.io/launch?region=ap-singapore)

09How we compare to GPU marketplaces

## The differences you can check yourself, line by line.

Same GPU, same hourly rate, a different bill. We charge by the second and include everything a GPU needs to run, and the example below shows what that changes on a real job.

### Same job, same hourly rate

A fine-tuning run of 2 h 14 m 03 s on an H100 SXM at $1.85 / h.

Billed per second

GPUBank

8,043 s

$4.13

Billed per started hour

Typical marketplace

46 min paid, not used

3 h

$5.55

**+ IPv4 address billed separately**

$0.00 here

**+ stopped for 2 days with a 200 GB disk**

$0.39 here, disk only

**+ a launch that fails its health check**

$0.00 here

$1.42 kept on this job

26 % of the per-hour bill, from counting seconds instead of hours.

20 such jobs a month

$28.40

Rates are equal on purpose: the whole difference comes from how time is counted.

[Pay for 8,043 seconds, not 3 hours](https://console.gpubank.io/launch?gpu=h100-sxm)

| Mechanism | GPUBank | Typical marketplace |
| --- | --- | --- |
| Billing granularity | per second | per minute or per hour |
| Charged while stopped | no, disk only | sometimes |
| IPv4 included | yes | often extra |
| CPU and RAM included | yes | often separate |
| Egress traffic | included | often metered |
| Health check before billing | yes | rare |
| Interruption notice | 5 min | varies |
| SLA covers all instance types | yes | often excludes spot |

Marketplace column: the public pricing and terms pages of twelve providers, read on 19 September 2026.

10Our numbers, published

## The figures you would ask us for are already on this page.

Launch time, health check pass rate, resume time, uptime per region and disk transfer speed, measured over the last 30 days, with our price history and every incident report next to them.

Median launch

48 s

p95 71 s

Health-check pass

99.2 %

first attempt

Resume time

31 s

same disk, same IP

Uptime in Paris

99.95 %

instance-hours

Transfer

1.6 GB/s

median, Paris to Frankfurt

### 90-day uptime by region

Daily instance availability, 2026-06-26 to 2026-09-23

Normal Storage incident

Paris

99.95 %

Frankfurt

99.97 %

Ashburn

99.96 %

Singapore

99.93 %

### H100 SXM price since launch

All-in price per hour, in USD

A price cut applies to running sessions at once; an increase is announced 30 days ahead and never touches a running session.

### Changelog

What changed and when, incidents included.

All systems operational

1. 2026-10-04 Model benchmarks Over 400 open models sized and priced on all ten GPUs, with the command that runs each one.
2. 2026-09-11 Per-second billing Every instance type is now billed by the second instead of the minute.
3. 2026-09-03 Incident report Storage latency in the EU regions for 41 minutes, with credits applied automatically.
4. 2026-08-27 Portable volumes Disks can move between regions from the console or the CLI.

11Contracts and security

## Every contract you sign, readable before you pay.

The terms, the privacy policy, the data processing agreement and the list of sub-processors are published in full and dated. You can read them without an account and without a sales call.

### Documents

Published in full, readable without an account.

- Service level and credits Availability target and automatic credits, for every instance type. 2026-10-04 [Read](https://gpubank.io/terms#availability)
- Data processing agreement For business customers, signed from the console, no sales call. 2026-10-04 [Read](https://gpubank.io/privacy#dpa)
- Sub-processors The public, dated list of who processes what. 2026-10-04 [Read](https://gpubank.io/privacy#sub-processors)
- Acceptable use policy What may and may not run on our GPUs. 2026-10-04 [Read](https://gpubank.io/terms#acceptable-use)
- Security Keys, encryption, isolation and access to your data. 2026-10-04 [Read](https://gpubank.io/privacy#security)

### Security

How your machines and your data are protected.

- SSH keys only, never passwords Instances accept key-based SSH. Password logins are disabled.
- Disks encrypted at rest with AES-256 Every persistent disk. Your own keys on request for reserved capacity.
- A whole GPU per instance, never shared You get the entire card, never a slice next to another customer.
- ISO 27001 certified partner datacenters EU data stays in the EU unless you move a disk yourself.

12What people rent GPUs for

## The right GPU for AI training, inference, image generation and research.

Most of our customers do one of four things, and each runs best on a different set of cards. The rules stay the same for all of them: per-second billing, a disk that outlives every instance and a usage line anyone can read.

Creators Run ComfyUI, SDXL, FLUX and AI video on cards from 24 to 96 GB, with your models kept on a disk that survives every stop. RTX 4090 RTX 5090 L40S RTX PRO 6000 From $0.23 / h

Fine-tuning Fine-tune an LLM with LoRA or QLoRA up to 70B, write checkpoints to your own disk and pick up where you left off after any interruption. A100 SXM H100 SXM H200 SXM From $0.95 / h

Inference Serve models behind an OpenAI-compatible API with warm workers, and check the real latency on the public status page. RTX PRO 6000 H100 SXM H200 SXM MI300X From $1.15 / h

Research Stretch a small research budget with per-second billing and a spend cap, and stop the GPU between experiments without losing your data. RTX 4090 L40S A100 SXM From $0.23 / h

13Frequently asked questions

## What people ask before renting their first GPU.

If your question is not answered here, the documentation covers the rest. Support tickets are reserved for customers and open from the console once your balance reaches $25, which is why we do not offer pre-sales support.

Yes. The hourly price of every GPU already covers the vCPUs, the RAM, the local NVMe scratch disk, one public IPv4 address and all traffic in and out. The only other line that can ever appear on your receipt is a persistent disk, at $0.03 per GB per month.

Only the persistent disk, for as long as you keep it. A stopped instance costs nothing for compute, IPv4 or scratch storage, and once you delete the disk nothing is billed at all.

The live board only lists cards you can actually launch, so you see a shortage before you try. You can then queue the launch until a card frees up, set an availability alert, or move your disk to a region that has stock.

Your instance is restarted on a healthy host with the same disk attached, the interrupted session is credited to your balance in full, and the incident is published on the status page with a post-mortem.

Yes, from the console or the CLI. The measured transfer speed is shown before you confirm, and a 200 GB disk usually moves between Paris and Frankfurt in about two minutes.

Exactly the one you choose. The H100 SXM and the H100 PCIe are sold as separate products with their own memory bandwidth and interconnect, and the console shows the PCI ID of the card you are running on.

GPUBank is run by GPU Compute OÜ, a company registered in Tallinn, Estonia, registry code 17984370. Our company details, service level, DPA and sub-processors are published for anyone to read.

You top up a prepaid balance in US dollars with crypto, starting at $25. Bitcoin, USDT, Ethereum and more than a dozen other coins are accepted, and larger deposits come with a bonus: a $50 top-up gives you $65 of credit, $100 gives you $135, $250 gives you $350, and from $1,000 the bonus reaches +60 %. The credit never expires. Usage is then deducted by the second, and every account can set a monthly spend cap.

Still a question?

### We are here once you are up and running.

Open a ticket from the console once your balance reaches $25. Over the last 30 days the median first reply was 22 minutes.

[Open the console](https://console.gpubank.io/)

[Support tickets in the console For customers with a balance of $25 or more](https://console.gpubank.io/support)

Full documentation Guides, examples and the API reference

All systems operational 99.95 % uptime over the last 30 days

Quick links

- [Read the docs](https://gpubank.io/docs)
- [Explore the API](https://gpubank.io/docs#reference)
- [Launch a GPU](https://console.gpubank.io/launch)

gpubank.io

```
$ gpubank launch \
    --gpu h100-sxm \
    --region eu-paris
● ready in 47.8 s
```

Four regions

- Paris
- Frankfurt
- Ashburn
- Singapore

Same hardware

The same GPUs, disks and network on every plan.

Get started from $25

## Launch your first GPU today and read exactly what it cost.

An H100 SXM in Paris costs $1.85 an hour, or $0.000514 / s, with everything included. Top up $50 in crypto and your balance shows $65, enough to run a real first job on a serious GPU.

[Create an account](https://console.gpubank.io/register)

Read the quickstart

- Bitcoin, USDT and a dozen other coins
- A spend cap on every account
- Stop at any second and keep your disk
- No minimum spend and no contract
- gpubank.io
  ```
  $ gpubank launch \
      --gpu h100-sxm \
      --region eu-paris
  ● ready in 47.8 s
  ```