Nebius AI Cloud review

Facts from Nebius AI Cloud’s own pricing and docs pages, September 2026 — 62 read from a quoted sentence on the page we archived.

Summary

Nebius AI Cloud rents GPUs from $1.80 per GPU-hour on demand, bills per GPU-hour. The minimum to start is $25 first payment. Egress is free. Nebius AI Cloud does not publish a refund policy.

GDPR · DPA99.50% uptime SLASpend limitsFree egress

From

$1.80/GPU-hr

Billing

Per GPU-hour

Minimum commitment

$25 first payment

Egress

Free

Uptime SLA

99.50% for Compute

Registered in

🇳🇱

Visit Nebius AI Cloud ↗

On-demand pricing

NVIDIA RTX PRO 6000 Blackwell 96GBLowest

×1, on demand · NVIDIA RTX PRO 6000

$1.80/hr

Deploy

NVIDIA H100 SXM5 80GB

×1, on demand · NVIDIA HGX H100

$4.50/hr

Deploy

NVIDIA H200 SXM5 141GB

×1, on demand · NVIDIA HGX H200

$5.40/hr

Deploy

NVIDIA B200 SXM6 180GB

×1, on demand · NVIDIA HGX B200

$8.50/hr

Deploy

NVIDIA B300 SXM6 288GB

×1, on demand · NVIDIA HGX B300

$9.50/hr

Deploy

See all 18 Nebius AI Cloud offers →

What’s good

Lowest published on-demand rate is $1.80/GPU-hr (NVIDIA RTX PRO 6000 Blackwell 96GB)

Egress is free — no per-GB charge on data leaving

Self-serve signup — no sales call to get started

Publishes a 99.50% for Compute uptime commitment, with service credits when it is missed

Free credits available: promo codes grant free credits

Commitment discounts: up to 35% off for multi-month cluster commitments

Plans

	On-demand	Preemptible	Committed
From	$1.55/GPU-hr (L40S) Nebius AI Cloud · On-demand Every on-demand rate we hold for Nebius AI Cloud, per GPU per hour, cheapest first. NVIDIA RTX PRO 6000 Blackwell 96GB ×1, on-demand · NVIDIA RTX PRO 6000 $1.80/hr Deploy NVIDIA H100 SXM5 80GB ×1, on-demand · NVIDIA HGX H100 $4.50/hr Deploy NVIDIA H200 SXM5 141GB ×1, on-demand · NVIDIA HGX H200 $5.40/hr Deploy NVIDIA B200 SXM6 180GB ×1, on-demand · NVIDIA HGX B200 $8.50/hr Deploy NVIDIA B300 SXM6 288GB ×1, on-demand · NVIDIA HGX B300 $9.50/hr Deploy Prices as last read from Nebius AI Cloud’s own pages. Open a card for every provider renting it.	$0.74 (L40S) Nebius AI Cloud · Preemptible Every spot rate we hold for Nebius AI Cloud, per GPU per hour, cheapest first. NVIDIA RTX PRO 6000 Blackwell 96GB ×1, spot · NVIDIA RTX PRO 6000 $1.80/hr Deploy NVIDIA H100 SXM5 80GB ×1, spot · NVIDIA HGX H100 $3.85/hr Deploy NVIDIA H200 SXM5 141GB ×1, spot · NVIDIA HGX H200 $4.50/hr Deploy NVIDIA B200 SXM6 180GB ×1, spot · NVIDIA HGX B200 $7.15/hr Deploy NVIDIA B300 SXM6 288GB ×1, spot · NVIDIA HGX B300 $7.85/hr Deploy Prices as last read from Nebius AI Cloud’s own pages. Open a card for every provider renting it.	Up to 35% below on-demand
Minimum term	None; $25 first payment	None	Multiple months
Billing	Per GPU-hour	Per GPU-hour	Usage commitment for multiple months
Uptime SLA	99.50% (Compute Cloud SLA)	—	99.50% (Compute Cloud SLA)
Tenancy	Non-virtualised GPU VMs, single node to clusters	Same platforms, interruptible	Reserved large-scale clusters
Provisioning	Self-serve in the console	Self-serve	Via sales
Payment	Bank card; bank transfer for companies	Bank card; bank transfer for companies	Companies only, via sales

Details

Billing & access

How you pay

Bank card or bank transfer; PDF invoices by email

Currency

USD, ex VAT

Commitment discounts

Up to 35% off for multi-month cluster commitments

Free credits

Promo codes grant free credits

Approval or KYC

Individuals or companies; commitments need a company account

What you get

GPUs listed

GB300 NVL72, HGX B300, GB200 NVL72, HGX B200, HGX H200, HGX H100, RTX PRO 6000, L40S

Regions

6 public regions: Finland, France, Israel, Kansas City (US), UK ×2

Plus private regions (Iceland, Madrid, France, Minnesota) for existing deployments. GPU platforms vary by region: H100 in Finland only; B300 in the UK, France and Minnesota; B200 in Israel and Kansas City; H200 in Finland, France, Kansas City and Iceland.

Node configuration

Per GPU: H100/H200 16 vCPU · 200 GB; B200 20 · 224; B300 24 · 346

Interconnect

NVLink in node; non-blocking Quantum-2 InfiniBand across clusters

Local storage

Local SSD block volumes $0.065/GiB-month

Persistent storage

Block $0.053–0.118/GiB-mo; shared filesystem $0.08; WEKA $0.10; object from $0.0147

Ingress

Free

Public endpoints

Public IP free

Private networking

Virtual Networks; Nebius Tunnels

Images

Boot disk images per platform

Access methods

Web console, CLI, API

Model deployment

Token Factory; Managed MLflow

Managed inference

Token Factory (managed inference)

Managed Kubernetes

Managed Kubernetes, free; managed Slurm (Soperator), free

Serverless

Serverless AI

Kubernetes integration

Managed Kubernetes; Soperator (Slurm); SkyPilot

FAQ

How much does Nebius AI Cloud cost?

As of September 2026, the lowest published on-demand rate is $1.80 per GPU-hour for the NVIDIA RTX PRO 6000 Blackwell 96GB. Nebius AI Cloud publishes an on-demand rate for 5 GPU models. Billing is per GPU-hour.

Does Nebius AI Cloud have an uptime SLA?

Nebius AI Cloud states 99.50% for Compute; compensation 10–30% of fees. Service credits: credit against future fees; claim within 14 days; sole remedy.

Is there a minimum commitment on Nebius AI Cloud?

Nebius AI Cloud states its minimum as $25 first payment; no term.

Does Nebius AI Cloud charge for egress?

No. Nebius AI Cloud states that egress is free.

Can I sign up to Nebius AI Cloud without approval?

Nebius AI Cloud states that signup is self-serve. On approval and identity checks it states: individuals or companies; commitments need a company account.

Where are Nebius AI Cloud's data centres?

Nebius AI Cloud lists 6 public regions: Finland, France, Israel, Kansas City (US), UK ×2.

Who do you contract with at Nebius AI Cloud?

The contracting entity is Nebius B.V..

Reviewed September 2026, from Nebius AI Cloud’s own site. 62 facts, each quoted from the page it was read on.