Observability and evaluation platform for AI agents and LLM apps, with tracing, issue detection and experiments on a free plan.
Freemium
Nebius AI Cloud
GPU cloud for AI training and inference with on-demand NVIDIA clusters, managed Kubernetes and Slurm, and usage-based billing.
What is Nebius AI Cloud?
Nebius AI Cloud is a cloud platform built for AI workloads that rents NVIDIA GPUs and supporting infrastructure by usage. Without approval or a sales call, you can launch up to 32 GPUs from the NVIDIA Hopper, Blackwell and Blackwell Ultra generations through the self-service console, on virtual machines that already have GPU and network drivers installed. Data centers are in Europe and the United States. You can orchestrate clusters with managed Kubernetes or Slurm, run one-click AI and ML applications, and manage everything through the console, API, CLI, Terraform or SDKs. Pricing is pay-as-you-go: on-demand rates include $5.40 per GPU-hour for HGX H200, $4.50 for HGX H100 and $1.80 for RTX PRO 6000, with cheaper preemptible spot capacity and commitment discounts for large reserved clusters. Storage, networking and managed services such as PostgreSQL and MLflow are billed separately, and some services in preview are free. Support is available around the clock through the console. To start, add $25 or more to your account.
Use cases
- Train machine learning models on GPUs
- Run AI inference in production
- Launch GPU clusters without a sales call
- Orchestrate jobs with Kubernetes or Slurm
- Reserve large clusters at discounted rates
- Run preemptible GPU jobs at lower cost
Who it's for
Why use Nebius AI Cloud
- You can start training on current NVIDIA GPUs without a waitlist or minimum commitment.
- Billing by the second means a 30-minute job on a GPU costs half the hourly rate.
- Managed Kubernetes and Slurm reduce the work of operating a cluster yourself.
- Preemptible capacity and commitment discounts give two ways to lower GPU spend.
- Round-the-clock support is available through the console.
Nebius AI Cloud features
- On-demand NVIDIA GPUs from L40S to HGX B300
- Self-service access to up to 32 GPUs
- Preemptible spot VMs with dynamic pricing
- Commitment discounts for multi-month clusters
- Managed Service for Kubernetes and Managed Soperator for Slurm
- Shared filesystem, object storage and block volumes
- GPU and InfiniBand monitoring dashboards, including Grafana
- Access through console, API, CLI, Terraform and SDKs
- Per-second billing for running compute
Nebius AI Cloud integrations
How to get started with Nebius AI Cloud
- Sign up for console access and add payment information; $25 or more lets you deploy GPU clusters right away.
- Choose a GPU platform and region, then launch VMs, or create a managed Kubernetes or Slurm cluster.
- Run training, inference or one-click AI and ML applications, and watch GPU metrics in the dashboards.
- Delete resources when finished to stop charges, or request commitments for stable long-running workloads.
Nebius AI Cloud limitations
- Self-service access is capped at 32 GPUs
- Spot prices change as often as every 15 minutes and can be interrupted
- Prices exclude taxes such as VAT, and Israeli companies are billed in shekels
- Large commitments require signing up as a company and contacting sales
Nebius AI Cloud pricing
Nebius AI Cloud has no free plan. Plans start at $0.06 usage-based, according to the official pricing page on .
Nebius AI Cloud plans
NVIDIA HGX B300
$9.50 per GPU-hour
On-demand
- Blackwell Ultra GPU
- 270 GB of GPU Memory
NVIDIA HGX B200
$8.50 per GPU-hour
On-demand
- Blackwell GPU
- 180 GB of GPU Memory
NVIDIA HGX H200
$5.40 per GPU-hour
On-demand
- Hopper GPU
- 141 GB of GPU Memory
- Preemptible from $0.79
NVIDIA HGX H100
$4.50 per GPU-hour
On-demand
- Hopper GPU
- 80 GB of GPU Memory
- Preemptible from $0.79
NVIDIA RTX PRO 6000
$1.80 per GPU-hour
On-demand
- Blackwell GPU
- 96 GB of GPU Memory
- Preemptible from $0.79
NVIDIA L40S with Intel CPU
from $1.55 per GPU-hour
On-demand
- Ada Lovelace GPU
- 48 GB of GPU Memory
CPU-only instances
from $0.06 per CPU-hour
On-demand
- Intel Ice Lake from $0.06
- AMD EPYC Genoa from $0.13
Plans as listed on the official pricing page on . Prices can change and may vary by country or exclude tax. Check the official pricing page.
Nebius AI Cloud alternatives
The main Nebius AI Cloud alternatives listed on Debuly are CoreWeave, Lambda, Runpod, Together AI, Vast.ai and Crusoe Cloud. The table compares pricing and platforms; order is A–Z.
| Tool | Pricing | Starting price | Platforms |
|---|---|---|---|
| Nebius AI Cloud | Paid | $0.06, usage-based | Web, API, CLI |
| CoreWeave | Paid | — | Web, API, Bare metal |
| Crusoe Cloud | Paid | — | Web, API, CLI |
| Lambda | Paid | $0.69, usage-based | Web, API, CLI |
| Runpod | Paid | — | Web, API |
| Together AI | Paid | — | Web, API |
| Vast.ai | Paid | — | Web, API, CLI |
CoreWeave
AI cloud that rents NVIDIA GPU capacity by the hour, with on-demand, spot and reserved plans for model training and inference.
Crusoe Cloud
AI cloud with on-demand GPU instances, managed Kubernetes and Slurm, and managed inference and fine-tuning for open models.
Lambda
GPU cloud for AI with on-demand NVIDIA instances and reserved 1-Click Clusters for training, fine-tuning and inference.
Runpod
Pay-per-second GPU cloud for AI: on-demand Pods, Serverless endpoints, multi-node Clusters and ready-made public model APIs.
Together AI
AI cloud for running open-source models through serverless and batch inference, dedicated endpoints, GPU clusters and fine-tuning.
Vast.ai
GPU cloud marketplace for renting on-demand, interruptible or reserved GPUs with per-second billing.
Nebius AI Cloud FAQ
How much do Nebius GPUs cost?
On-demand prices per GPU-hour are $9.50 for HGX B300, $8.50 for HGX B200, $5.40 for HGX H200, $4.50 for HGX H100, $1.80 for RTX PRO 6000 and from $1.55 for L40S with Intel CPUs.
Does Nebius offer cheaper spot GPUs?
Yes. Preemptible instances start from $0.79 per GPU-hour on H100, H200 and RTX PRO 6000. Spot prices are dynamic and can change as often as every 15 minutes.
Is there a minimum spend or free tier?
There is no listed free tier. Self-service GPU clusters become available once you add $25 or more to your account, and there are no minimum commitments for on-demand use.
How can I get a discount on Nebius?
Usage commitments help for long-term, stable workloads. Sign up as a company and contact the sales team to request them, since large clusters can be reserved for multiple months.
How do I access Nebius AI Cloud?
Use the web console, or connect through the API, CLI, Terraform or SDKs. Self-service gives access to up to 32 GPUs in data centers in Europe and the United States.
What are alternatives to Nebius?
Alternatives include CoreWeave, Lambda, RunPod, Together AI and the GPU offerings from AWS, Google Cloud and Azure. Compare GPU availability, hourly rates, networking and support.
Sources & verification
Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.
| Source | Type | Accessed |
|---|---|---|
| nebius.com/ | official | |
| docs.nebius.com/applications/standalone/pricing | pricing | |
| nebius.com/prices | pricing | |
| docs.nebius.com/signup-billing | pricing | |
| docs.nebius.com/signup-billing/billing-models/payg | pricing | |
| nebius.com/self-service | official |
Image credits
- Logo: from nebius.com · fetched · Vendor trademark, used only to identify the product; removed on request
- Cover: none collected yet, so a generated tile is shown.
Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.
More in Developer platforms 4
All developer platforms toolsDeployment platform for Node.js, Python and AI apps with Git deploys, one-click templates, an AI copilot and an MCP server for agents.
Free trialEvaluation platform for testing LLM applications with test suites, plus custom benchmarks built from your GitHub repo.
GPU cloud marketplace for renting on-demand, interruptible or reserved GPUs with per-second billing.
Paid
