There is a discount from 10% to 50% on all GPU models for an advance payment of 1-6 months.

Products

Cloud servers

Cloud platform with the latest GPUs, fast onboarding, per‑second billing, and immersion cooling. Isolated resources ensure maximum performance for your project..

GPU servers

Cloud servers with modern RTX and Tesla graphics accelerators for games, rendering, streaming, working with 3D graphics, and artificial intelligence.

H200

H100 NVL

H100

RTX 5090

RTX 4090

RTX 3090

RTX 3080

A100

A10

A2

RTX 2080 Ti^EOL

Tesla T4^EOL

Tesla V100^EOL

All GPU servers

CPU servers

The cloud servers with high-performance Intel Xeon Gold 2nd, 3rd and 5th generation CPU are available for 100% of the processor time.
SSD servers до 75К IOPS NVMe servers до 360К IOPS
All CPU servers

Immers Foundation Models

The largest catalog of vetted open‑source models with automatic configuration selection and tuning for rapid deployment. Launch private endpoints with no token fees, or use public endpoints.

GLM-5.3
Qwen3.8-2.4T-A95B
Qwen3.8-27B
DeepSeek-V4-Flash-0731
Kimi-K3
Unlimited-OCR
GLM-5.2
Kimi-K2.7-Code
NVIDIA-Nemotron-3-Ultra-550B-A55B
gemma-4-12B-it
MiniMax-M3
PaddleOCR-VL-1.6
DeepSeek-V4-Pro
DeepSeek-V4-Flash
Qwen3.6-27B
Qwen3.6-35B-A3B
Kimi-K2.6
GLM-5.1
gemma-4-26B-A4B-it
gemma-4-31B-it
NVIDIA-Nemotron-3-Super-120B-A12B
Qwen3.5-122B-A10B
Qwen3.5-397B-A17B
gpt-oss-120b
gpt-oss-20b

All models from catalogue

Dedicated servers

Rent a physically dedicated server for a long term with a monthly payment. Configure it using modern components: Intel Xeon Gold 2nd to 5th generation processors and up to 8192 GB of RAM per server, SSD and NVMe disks for data centers.

Select a dedicated server до 10 GPU и 2.5M IOPS

Marketplace

Use popular and modern applications as effective tools for organizing your project. Save time with pre-configured images that already have all the necessary components installed.

Forget about manually downloading and installing the software — just deploy a virtual server with a ready-made image.

Neural networks
3D
CUDA
Docker / NGC
For games
Windows images
Linux images

All pre-installed images
Features
Prices
FAQ
Contact
Login

Cloud Resource and Service Pricing

CPU

Option	Price, hour	Price, month
vCPU, per core¹
RAM, per GB

CPU servers

GPU

Prices:

Option	Price
H200 141GB PCI-E
H100 NVL 94GB PCI-E
H100 80GB PCI-E
RTX 5090 32GB PCI-E
RTX 4090 24GB PCI-E
RTX 3090 24GB PCI-E
RTX 3080 10GB PCI-E
A100 80GB PCI-E
A10 24GB PCI-E
A2 16GB PCI-E
RTX 2080 Ti 11GB PCI-E
Tesla T4 16GB PCI-E
Tesla V100 32GB PCI-E

GPU servers

Storage

Option	Price, hour	Price, month
SATA SSD (Local), per GB²
NVMe SSD (Local), per GB²
Volume (Volume-backed), per GB	$0.00003
Volume SSD (Volume-backed), per GB	$0.00014
Snapshot, per GB	$0.00003
Volume Snapshot, per GB	Free

Network

Option	Price, hour	Price, month
IP-address

Operating systems

Option	Price, hour	Price, month
Windows Server 2019 Standard
Windows Server 2019 Datacenter edition, per core
Windows Server 2022 Standard
Windows Server 2022 Datacenter edition, per core
Windows Server 2025 Standard
Windows Server 2025 Datacenter edition, per core

Isolated virtual core of Intel® Xeon® Scalable 2-nd, 3-rd and 5th generation. Without overselling.
SATA SSD is the default option. To setup an NVMe instance choose Local instance and one of the NVMe configuration during Virtual Machine creation.

Tokens

Model	Price, input/1M	Price, output/1M	Price, cache/1M
Qwen3.8-27B	$0.269	$1.343	$0.056
DeepSeek-V4-Flash-0731	$0.078	$0.336	$0.022
Qwen3.6-27B	$0.627	$3.918	$0.112
Qwen3.6-35B-A3B	$0.078	$0.448	$0.011
gemma-4-26B-A4B-it	$0.084	$0.470	$0.008
Qwen3-Coder-Next	$0.313	$1.567	$0.157
gpt-oss-20b	$0.034	$0.157	$0.045

Listed prices apply to public endpoints hosted on servers and virtual machines located within immers.cloud’s own infrastructure.
Caching is enabled by default for all public endpoints, but its availability is not guaranteed.
Caching applies exclusively to input tokens.
These prices will take effect on 01.10.2026. Until that date, you may use public endpoints with the listed models free of charge.