What you can do with this service
AI training and inference
Modeling and design
Virtual Graphical Workstations
Visualization and rendering
Video processing
AR/VR and simulations
Pricing
Per hour
Per month
Per month
73 611 UZS
53 000 000 UZS
53 000 000 UZS
Large LLMs (70B+), FP8, Transformer Engine, HBM3
AI training (evaluation)
989 TFLOPS
FP16
80 GB HBM3
240
RAM
40
CPU
H100
Per hour
Per month
Per month
57 175 UZS
41 166 000 UZS
41 166 000 UZS
Standard for LLM, NVLink, high bandwidth
AI training (evaluation)
312 TFLOPS
FP16
80 GB HBM2e
120
RAM
20
CPU
A100
Per hour
Per month
Per month
47 485 UZS
34 190 000 UZS
34 190 000 UZS
Good for single-node training of medium-sized LLMs
AI training (evaluation)
362 TFLOPS
FP16
48 GB GDDR6
120
RAM
20
CPU
L40s
Per hour
Per month
Per month
10 440 UZS
7 500 000 UZS
7 500 000 UZS
Medium models, without NVLink
AI training (evaluation)
77 TFLOPS
FP16
16 GB GDDR6
100
RAM
26
CPU
RTX A4000
Per hour
Per month
Per month
6 948 UZS
5 002 800 UZS
5 002 800 UZS
Small models / fine-tuning
AI training (evaluation)
65 TFLOPS
FP16
16 GB GDDR6
48
RAM
8
CPU
T4
Operating system images
Platform
vGPU virtual machines run on the Elastic Cloud platform
Fault-tolerant infrastructure
SSD storage
Resource isolation
Internet 100Mbps
NVIDIA GPU Support
vGPU delivers high performance for graphics, ML, and video workloads. It supports CUDA, RTX, and hardware acceleration.
It takes 5-7 minutes to create a virtual machine with a GPU.
You choose the CPU, RAM, and GPU configuration, and the system automatically places the virtual machine on the optimal node.
You choose the CPU, RAM, and GPU configuration, and the system automatically places the virtual machine on the optimal node.
Quick start
Performance
- Full self-service
- Web interface
- API
- Terraform
Control panel
Backups
Create snapshots before updates and roll back quickly if necessary.
Built-in DR backup service. Choose the desired storage period: from 2 to 30 days.
Migration to Cloupard
Want to move your project to the cloud? We’ll make the migration painless. We’ll move your infrastructure, configure the environment, and verify that everything works as expected. You can continue working while we take care of the migration.
FAQ
Yes. vGPU-powered virtual machines deliver high performance for both model training and inference, and efficiently handle large-scale data processing.
Pricing depends on your CPU, RAM, and GPU configuration and is calculated in the configurator.
You can deploy a VM on either Linux or Windows. Both platforms ship with full support for NVIDIA drivers and GPU tooling.
Yes. RTX A4000 configurations are ideal for designers, engineers, and architects working with 3D graphics, modeling, and visualization.
You can create, configure, and manage virtual machines using the management console.
Yes. Virtual machines support connection through corporate VPN, Active Directory, and internal cloud networks, ensuring secure integration with your organization's infrastructure.
NVIDIA Tesla T4, RTX A6000, Tesla V100, and other GPUs are available on request.
Сообщение об успешной отправке!
