Commitment discounts
A committed volume of services (CVoS) is an agreement to receive a guaranteed discount for consuming a certain amount of Nebius AI services within 3 months, 6 months, or 1 year.
A base pay-as-you-go price for NVIDIA® Tensor Core H100 GPU is $3.5 per hour. But if you’re sure that your consumption of the Nebius AI services is going to be relatively stable and predictable, you can save significantly with a guaranteed discount.
Clusters tailored for LLM-scale training
Let’s say you’re training a large language model. These cluster configurations — customizable, of course — will work for you. For example, 512 H100 GPUs reserved for a month will let you train the Chinchilla-scale model with 70 billion parameters.
$2.12/h
Price per 1 GPU H100 SXM5 + InfiniBand.
3-year reserve*
Configuration
- 64 × 8 NVIDIA® H100
- 64 × 160 Intel® Sapphire Rapids vCPUs
- 64 × 1.25 TB RAM
- 128 TB Network SSD
Estimated total price for the configuration if running 24/7 is $28,855,664.
$2.94/h
Price per 1 H100 SXM5 GPU + InfiniBand.
1-year reserve
The complete configuration:
- 64 × 8 NVIDIA® H100
- 64 × 160 Intel® Sapphire Rapids vCPUs
- 64 × 1.25 TB RAM
- 128 TB Network SSD
Estimated total price for the configuration if running 24/7 is $13,296,353.
$3.2/h
Price per 1 H100 SXM5 GPU + InfiniBand.
6-month reserve
The complete configuration:
- 64 × 8 NVIDIA® H100
- 64 × 160 Intel® Sapphire Rapids vCPUs
- 64 × 1.25 TB RAM
- 128 TB Network SSD
Estimated total price for the configuration if running 24/7 is $7,231,242.
$3.34/h
Price per 1 H100 SXM5 GPU + InfiniBand.
3-month reserve
The complete configuration:
- 64 × 8 NVIDIA® H100
- 64 × 160 Intel® Sapphire Rapids vCPUs
- 64 × 1.25 TB RAM
- 128 TB Network SSD
Estimated total price for the configuration if running 24/7 is $3,772,600.
Nodes for stable diffusion fine-tuning
Alternatively, a specialized node would be a completely different beast if your flow requires continuous fine-tuning. Refine your model easily with these offers.
$2.12/h
Price per 1 GPU H100 SXM5 + InfiniBand.
3-year reserve*
Configuration
- 1 × NVIDIA® H100
- 1 × 20 Intel® Sapphire Rapids vCPUs
- 1 × 160 GB RAM
- 1 × 200 GB SSD
Estimated total price for the configuration if running 24/7 is $56,217.
$2.94/h
Price per 1 H100 SXM5 GPU + InfiniBand.
1-year reserve
The complete configuration:
- 1 × NVIDIA® H100
- 1 × 20 Intel® Sapphire Rapids vCPUs
- 1 × 160 GB RAM
- 1 × 200 GB SSD
Estimated total price for the configuration if running 24/7 is $25,922.
$3.2/h
Price per 1 H100 SXM5 GPU + InfiniBand.
6-month reserve
The complete configuration:
- 1 × NVIDIA® H100
- 1 × 20 Intel® Sapphire Rapids vCPUs
- 1 × 160 GB RAM
- 1 × 200 GB SSD
Estimated total price for the configuration if running 24/7 is $14,100.
$3.34/h
Price per 1 H100 SXM5 GPU + InfiniBand.
3-month reserve
The complete configuration:
- 1 × NVIDIA® H100
- 1 × 20 Intel® Sapphire Rapids vCPUs
- 1 × 160 GB RAM
- 1 × 200 GB SSD
Estimated total price for the configuration if running 24/7 is $7,357.
Any data‑intensive case covered
Whether it’s learning, fine-tuning, inference or any other set of tasks, our experts will come up with the right cluster configuration and an affordable price for you.
The provided information and prices do not constitute an offer or invitation to make offers or invitation to buy, sell or otherwise use any services, products and/or resources referred to on this website and may be changed by Nebius at any time. Contact sales to get a personalized offer.
All prices are shown without any applicable taxes, including VAT.
✻ The price is valid for a 3-year reserve with 50% prepayment.