Skip to content
Cloudj
GPU Cloud

GPU cloud: rent NVIDIA GPU servers by the hour

Rent NVIDIA GPU servers for LLM inference, fine-tuning and image generation. Live prices, per-minute billing from your balance, Ubuntu + CUDA and PyTorch + JupyterLab templates, SSH key access.

Rent a GPU serverSee live GPU prices
  • Live inventory, price confirmed before you order
  • Filter machines by country with two-letter codes
  • A stopped instance is billed for storage only

With Cloudj GPU Cloud you rent a GPU server from the live inventory: choose the GPU model, count and disk, filter by country when location matters, start from the Ubuntu 22.04 + CUDA or PyTorch + JupyterLab template, then connect over SSH with your key. Usage is billed per minute from your account balance, and a stopped instance is only billed for its storage.

GPU rental prices and live availability

Live inventory: availability and prices change with the market. The console confirms the final price before you order.

GPUVRAM per GPUArchitectureFP32 computePower limitFrom
NVIDIA RTX PRO 6000 Max-Q 96 GBBlackwell≈ 119 TFLOPS300 WFrom $1.62/hDetails →
NVIDIA RTX PRO 6000 WS 96 GBBlackwell≈ 119 TFLOPS600 WFrom $1.77/hDetails →
NVIDIA RTX 5090 32 GBBlackwell≈ 107.6 TFLOPS575 WFrom $0.70/hDetails →
NVIDIA RTX 5090D 32 GBBlackwell≈ 107.6 TFLOPS575 WOut of stockDetails →
NVIDIA RTX PRO 6000 S 96 GBBlackwell≈ 93.6 TFLOPS600 WFrom $1.29/hDetails →
NVIDIA RTX 4090 24 GBAda Lovelace≈ 81.4 TFLOPS450 WFrom $0.49/hDetails →
NVIDIA RTX 6000Ada 48 GBAda Lovelace≈ 81.4 TFLOPS300 WFrom $0.98/hDetails →
NVIDIA RTX 6000D 84 GBBlackwell≈ 77.6 TFLOPS600 WOut of stockDetails →
NVIDIA L40S datacenter48 GBAda Lovelace≈ 73.3 TFLOPS350 WOut of stockDetails →
NVIDIA RTX PRO 5000 48 GBBlackwell≈ 69.6 TFLOPS300 WFrom $0.76/hDetails →
NVIDIA H100 SXM datacenter80 GBHopper≈ 53.5 TFLOPS700 WOut of stockDetails →
NVIDIA RTX 5080 16 GBBlackwell≈ 53.4 TFLOPS360 WFrom $0.49/hDetails →
NVIDIA RTX 4080S 16 GBAda Lovelace≈ 50.9 TFLOPS320 WOut of stockDetails →
NVIDIA RTX 4080 16 GBAda Lovelace≈ 48.6 TFLOPS285 WOut of stockDetails →
NVIDIA H100 NVL datacenter94 GBHopper≈ 48.3 TFLOPS400 WFrom $3.57/hDetails →
NVIDIA H200 NVL datacenter141 GBHopper≈ 48.3 TFLOPS600 WFrom $4.38/hDetails →
NVIDIA RTX 5070 Ti 16 GBBlackwell≈ 44.3 TFLOPS300 WOut of stockDetails →
NVIDIA RTX PRO 4000 24 GBBlackwell≈ 44.3 TFLOPS145 WFrom $0.26/hDetails →
NVIDIA RTX 4070S Ti 16 GBAda Lovelace≈ 42.6 TFLOPS285 WFrom $0.22/hDetails →
NVIDIA H100 PCIE datacenter80 GBHopper≈ 41 TFLOPS350 WOut of stockDetails →
NVIDIA RTX 4500Ada 24 GBAda Lovelace≈ 38.2 TFLOPS210 WFrom $0.41/hDetails →
NVIDIA RTX 3090 Ti 24 GBAmpere≈ 36.4 TFLOPS450 WOut of stockDetails →
NVIDIA RTX A6000 48 GBAmpere≈ 36.1 TFLOPS300 WFrom $0.53/hDetails →
NVIDIA RTX 4070S 12 GBAda Lovelace≈ 35.8 TFLOPS220 WOut of stockDetails →
NVIDIA RTX 3090 24 GBAmpere≈ 35.4 TFLOPS350 WFrom $0.26/hDetails →
NVIDIA RTX 3080 Ti 12 GBAmpere≈ 35.1 TFLOPS350 WFrom $0.16/hDetails →
NVIDIA A40 datacenter48 GBAmpere≈ 29.9 TFLOPS300 WFrom $0.59/hDetails →
NVIDIA RTX 3080 10 GBAmpere≈ 29.5 TFLOPS380 WFrom $0.19/hDetails →
NVIDIA RTX 4070 12 GBAda Lovelace≈ 29.3 TFLOPS200 WOut of stockDetails →
NVIDIA L4 datacenter24 GBAda Lovelace≈ 24.2 TFLOPS72 WFrom $0.42/hDetails →
NVIDIA RTX 5060 Ti 16 GBBlackwell≈ 22.8 TFLOPS180 WFrom $0.17/hDetails →
NVIDIA RTX 4060 Ti 16 GBAda Lovelace≈ 21.6 TFLOPS165 WFrom $0.14/hDetails →
NVIDIA RTX A4000 16 GBAmpere≈ 20.6 TFLOPS140 WFrom $0.11/hDetails →
NVIDIA RTX 3070 8 GBAmpere≈ 19.8 TFLOPS220 WOut of stockDetails →
NVIDIA RTX 3060 Ti 8 GBAmpere≈ 15.7 TFLOPS200 WFrom $0.09/hDetails →
NVIDIA A100 PCIE datacenter40 GBAmpere≈ 15.6 TFLOPS250 WFrom $0.50/hDetails →
NVIDIA A100 SXM4 datacenter40 GBAmpere≈ 15.6 TFLOPS400 WFrom $0.81/hDetails →
NVIDIA A800 PCIE datacenter80 GBAmpere≈ 15.6 TFLOPS300 WOut of stockDetails →
NVIDIA RTX 2080 Ti 11 GBTuring≈ 14.6 TFLOPS250 WFrom $0.10/hDetails →
NVIDIA RTX 3060 laptop 12 GBAmpere≈ 12.9 TFLOPS80 WOut of stockDetails →
NVIDIA Tesla V100 datacenter16 GBVolta≈ 12.5 TFLOPS300 WFrom $0.15/hDetails →
NVIDIA CMP 170HX datacenter64 GBAmpere≈ 12.1 TFLOPS250 WFrom $0.43/hDetails →
NVIDIA RTX 3060 12 GBAmpere≈ 12 TFLOPS170 WFrom $0.09/hDetails →
NVIDIA Titan Xp 12 GBPascal≈ 11.7 TFLOPS160 WOut of stockDetails →
NVIDIA RTX A2000 6 GBAmpere≈ 11.2 TFLOPS70 WFrom $0.07/hDetails →
NVIDIA GTX 1080 Ti 11 GBPascal≈ 11 TFLOPS250 WOut of stockDetails →
NVIDIA GTX 1080 8 GBPascal≈ 7.8 TFLOPS130 WOut of stockDetails →
NVIDIA GTX 1070 Ti 8 GBPascal≈ 7.4 TFLOPS180 WOut of stockDetails →
NVIDIA GTX 1650 4 GB≈ undefined TFLOPS WOut of stockDetails →
NVIDIA Q RTX 8000 48 GBTuring≈ 11.9 TFLOPS250 WFrom $0.27/hDetails →
NVIDIA Tesla P40 datacenter24 GBPascal≈ 9.4 TFLOPS250 WOut of stockDetails →
NVIDIA RTX 2070S 8 GBTuring≈ 8.8 TFLOPS210 WFrom $0.15/hDetails →
NVIDIA Q RTX 4000 8 GBTuring≈ 7.7 TFLOPS125 WFrom $0.13/hDetails →
NVIDIA RTX 2070 8 GBTuring≈ 7.7 TFLOPS150 WOut of stockDetails →
NVIDIA RTX 2060S 8 GBTuring≈ 7.5 TFLOPS175 WOut of stockDetails →
NVIDIA GTX TITAN X 12 GBMaxwell≈ 6.8 TFLOPS150 WOut of stockDetails →
NVIDIA RTX 2060 12 GBTuring≈ 6.5 TFLOPS130 WOut of stockDetails →
NVIDIA GTX 1660 Ti 6 GBTuring≈ 5.3 TFLOPS120 WOut of stockDetails →
NVIDIA GTX 1660 S 6 GBTuring≈ 3.6 TFLOPS125 WOut of stockDetails →

What you can run on a rented GPU

LLM inference

Install vLLM, SGLang, Ollama or llama.cpp over SSH, or bring them in your own Docker image, to serve and test language models.

Fine-tuning

Fine-tune language models on your own data with PyTorch and libraries such as Unsloth or PEFT that you install yourself.

Image and video generation

Install ComfyUI or other diffusion tools for Flux, SDXL and video workflows.

PyTorch and CUDA development

Start from the Ubuntu 22.04 + CUDA or PyTorch + JupyterLab template and work over SSH or in JupyterLab.

Headless rendering and video

Render with Blender or encode with FFmpeg from the command line; GPU instances have no graphical desktop.

How to rent a GPU server in 4 steps

  1. 01

    Choose a GPU model and, if location matters, the countries to search

  2. 02

    Pick a machine from the live list, then a template and disk size

  3. 03

    Select your SSH key, confirm the price and deploy from your account balance

  4. 04

    Connect over SSH with the host and port shown on the instance page

GPU cloud pricing and billing

  • Per minute: a running instance is charged for the GPU and its disk from your account balance; a stopped instance is charged for storage only.
  • Bandwidth is charged per GB downloaded and uploaded, at the price listed for the machine.
  • The price shown when you deploy is locked for that instance. GPU instances have no monthly or yearly plans.
  • Prices are in USD; you can display them in VND at the current exchange rate.

Guides and how-tos

GPU cloud Asia guide: Japan, Singapore, Korea and nearbyA practical GPU cloud Asia guide: which country codes to search for East and Southeast Asian users, a fallback order, and how to pair a GPU with a VPS.GPU cloud by location: how to pick a country for your GPUHow GPU cloud by location works on Cloudj: filter a live list of machines by country code, decide between near users, lowest price or data rules.GPU cloud Europe: choosing a country for your AI workloadsA practical guide to GPU cloud Europe searches: filter by several country codes at once, weigh Western, Nordic and Eastern hosts, and plan where data lives.GPU cloud North America: renting GPUs in the US and CanadaA GPU cloud North America guide: how to search US and CA hosts, read the results, pick a VPS front end and decide when a US GPU suits users elsewhere.

All articles →

GPU cloud FAQ

Pair your GPU server with

Cloud servers for websites, APIs and databases

→

S3 storage for datasets, checkpoints and backups

→

Full price list

→

API documentation for automation

→

GPU cloud: rent NVIDIA GPU servers by the hour

Rent a GPU serverRent an RTX 5090