GPUBank vs Scaleway
Scaleway is the French cloud subsidiary of the iliad Group, renting NVIDIA L4, L40S, H100 and B300 instances by the minute in Paris and Warsaw, with managed Kubernetes and managed inference. Its H100 SXM instances cost EUR 25.33 per hour for 8 GPUs, about $3.63 per GPU-hour converted and excluding VAT, and up to $3.79 on the 2-GPU size; GPUBank charges $1.85. Scaleway offers EU sovereignty, HDS and ISO 27001 certifications and managed Kubernetes; GPUBank costs about half and takes crypto.
GPUBank
Tallinn, Estonia
H100 SXM
$1.85 / h
- vCPU, RAM, NVMe, IPv4 and traffic included
- Per-second billing after the health check
- Top up in Bitcoin, USDT and 12+ other coins
vCPU, RAM, NVMe, IPv4 and traffic included
Scaleway
Paris, France
H100 SXM
$3.63 / h
- HDS health-data certification and ISO/IEC 27001:2022
- Free 24/7 support tickets for every account, paid plans with a hotline
- Kapsule managed Kubernetes and managed inference
Scaleway
European cloud
France28 points compared15 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
EU sovereignty, HDS and managed services at Scaleway. Half the price and crypto top-ups at GPUBank.
Scaleway fits European companies that need HDS-certified hosting in France, EUR invoices with VAT, managed Kubernetes and managed inference. For the GPUs both of us rent, GPUBank costs 49 to 67 % less per GPU-hour, and it includes IPv4 where Scaleway bills IPv4 and block storage separately. GPUBank opens with an email and password, bills by the second and runs in four regions on three continents.
ScalewayGPUBank 01The real cost
The same GPUs, side by side, per hour.
Every GPU model both of us rent, at each provider's own published price for one card, in US dollars, excluding VAT. Prices last checked on 29 September 2026.
| GPU | Scaleway per hour | GPUBank per hour | Difference | 100 h at Scaleway | 100 h at GPUBank |
|---|---|---|---|---|---|
 H100 SXM80 GB |
$3.63 | $1.85 | −49 % | $363.00 | $185.00 |
 H100 PCIe80 GB |
$3.28 | $1.35 | −59 % | $328.00 | $135.00 |
 L40S48 GB |
$1.68 | $0.55 | −67 % | $168.00 | $55.00 |
Scaleway prices are in EUR excluding VAT, converted at 1 EUR = 1.1460 USD. The H100 SXM is the 8-GPU instance rate and the H100 PCIe the 2-GPU rate, divided per GPU.
A fine-tune, 40 h on an H100 SXM
Scaleway
$145.20
GPUBank
$74.00
Saves $71.20 at GPUBank (−49 %)
An inference box, a month, 720 h, on an L40S
Scaleway
$1,209.60
GPUBank
$396.00
Saves $813.60 at GPUBank (−67 %)
A training run, 100 h on an H100 PCIe
Scaleway
$328.00
GPUBank
$135.00
Saves $193.00 at GPUBank (−59 %)
02Their catalogue
Every GPU Scaleway lists.
Scaleway lists 6 GPU models. The 5 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 3 of them are also rented at GPUBank (in blue).
B300288 GB · on demand$8.60- H100 SXM80 GB · on demand$3.63
- H100 PCIe80 GB · on demand$3.28
- L40S48 GB · on demand$1.68
L424 GB · on demand$0.90At GPUBank
vCPU, RAM, NVMe and traffic included. All prices
03What the price includes
What appears on the bill besides the GPU.
The same H100 SXM session of 2 h 14 m 03 s, line by line, as each provider describes its billing.
Scaleway
2 h 14 m 03 s on H100 SXM
- GPU · $3.63 / h
- $8.11
- vCPU and RAM
- Included
- Storage
- About $0.11 per GB-month (Block 5K)
- Egress
- Included for Instances
- Public IPv4
- EUR 0.005 per hour per flexible IPv4
- Billing granularity
- Per minute
- Minimum charge
- 60 minutes per resource (billing FAQ)
GPUBank
2 h 14 m 03 s on H100 SXM
- GPU · $1.85 / h
- $4.13
- vCPU and RAM
- Included
- Storage
- Local NVMe included; persistent disk $0.03 per GB-month
- Egress
- Included
- Public IPv4
- Included
- Billing granularity
- Per second, from the health check
- Minimum charge
- None
While stopped
What keeps running on the bill.
Scaleway
Compute stops; volumes, snapshots and IPs billed.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
2 countries against 4, and where they are.
Countries where Scaleway says its GPUs run, from its own region list or documentation, against GPUBank's four regions. A pin marks a country, not the exact site.
France
Poland
Paris
Frankfurt
Ashburn
Singapore
Scaleway2
- France
Poland
GPUBank4
- Paris
Frankfurt
Ashburn
Singapore
Scaleway · 2 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Scaleway
Payments and sign-up
Payment methods
No cryptoMinimum top-up Not published
To start
- 1Email and password
- 2Card
- 3Support request for multi-GPU and H100/B300 types
Billing modelPostpaid monthly invoice
GPUBank
Payments and sign-up
Payment methods
A prepaid USD balance, topped up in crypto. Every GPU-second is drawn from it.
Minimum top-up $25
To start
- 1Email and password
- 2Top up in crypto
- 3Launch a GPU
Billing modelPrepaid balance
06Trust and support
Certificates, uptime and who answers.
| Scaleway | GPUBank | |
|---|---|---|
| SOC 2 | Not published | Not published |
| ISO 27001 | Yes | Not published |
| HIPAA | Not published | No |
| GDPR DPA | DPA available | Published DPA, applies automatically |
| Availability target | 99.5 % on GPU Instances | 99.9 % monthly, per region |
| Status page | status.scaleway.com | Public status page |
| Support | Free 24/7 tickets; paid plans from EUR 50/month | Tickets from a $25 balance |
07Their site, our site
What you see before you pay.
Both sites as a visitor sees them, without an account: the Scaleway pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1The GPU pricing page shows EUR per hour and an approximate monthly figure, excluding VAT, per zone selector.
- 2The default zone, PAR-1, shows most GPU types as 'Not available in this zone'.
- 3The cost estimator on the GPU page adds block volume and flexible IPv4 lines to the instance price.
- 1Live stock per region and every price, without an account.
- 2Per-hour and per-second prices, with vCPU, RAM, NVMe, IPv4 and traffic in them.
- 3A machine-readable price feed at /prices.json.
08The spec sheet
28 points, read off both sites.
GPUBank aheadScaleway aheadEven
| Criterion | GPUBank | Scaleway |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourH100 SXM is priced only in 2, 4 or 8-GPU instances; $3.63 is the 8-GPU rate. | $1.85, all-in | $3.63 to $3.79, EUR converted, excl. VAT |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.90, L4, excl. VAT |
| GPU models listed | 10 | 6 |
| What the GPU price includesA GPU OS system volume needs at least 125 GB of block storage. | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, scratch NVMe; block volume and IPv4 extra |
| Billing | ||
| Billing granularity | Per second | Per minute |
| Minimum chargeThe Instances FAQ says GPU Instances are billed per minute of uptime; both statements are published. | None | 60 minutes per resource (billing FAQ) |
| Stopped instance | Disk only | Compute stops; volumes, snapshots and IPs billed |
| Persistent storage | $0.03 per GB-month | About $0.11 per GB-month (Block 5K) |
| Egress traffic | Included | Included for Instances |
| Public IPv4 | Included | EUR 0.005 per hour per flexible IPv4 |
| Billing model | Prepaid balance | Postpaid monthly invoice |
| Payment and sign-up | ||
| Crypto payments | Bitcoin, USDT, Ethereum and 12+ more | No |
| Card paymentsA physical card is required even with SEPA; virtual and prepaid cards are refused. | No | Visa, MasterCard, CB, Amex, UnionPay; SEPAAhead |
| Minimum deposit | $25 | Not published |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | API and scw CLI |
| Terraform | Official provider | Official provider |
| Serverless GPUs | No | Serverless Generative APIs, managed inferenceAhead |
| Managed Kubernetes | No | Kapsule managed, with GPU operatorAhead |
| Ready-made templates | 25 | GPU OS image with Docker and NVIDIA toolkit |
| Interruptible or spot | Yes, 40 % off | No |
| GPUs per instance | Up to 8 | Up to 8; DGX clusters of 16 to 1,016 GPUsAhead |
| Multi-node interconnect | Not offered | NVLink 900 GB/s; InfiniBand on clustersAhead |
| Network and regions | ||
| RegionsH100 SXM and B300 are in PAR-2 only. | 4 | PAR-1, PAR-2, WAW-2 |
| Countries | France, Germany, US, Singapore | France, Poland |
| EU regions | Paris, Frankfurt | Paris and Warsaw |
| Trust and support | ||
| Availability targetGPUBank's terms set a 99.9 % monthly target per region. | 99.9 % monthly | 99.5 % on GPU Instances |
| Compliance reports | None published | ISO/IEC 27001:2022, HDS; SOC 2 not publishedAhead |
| Public status page | Yes | status.scaleway.com |
| Support | Tickets from a $25 balance | Free 24/7 tickets; paid plans from EUR 50/monthAhead |
09The right fit
Different needs. A clear choice.
What Scaleway does better
- HDS health-data certification and ISO/IEC 27001:2022
- Free 24/7 support tickets for every account, paid plans with a hotline
- Kapsule managed Kubernetes and managed inference
- B300 and 8-GPU H100 SXM with NVLink, DGX clusters on request
- L4 from $0.90 per GPU-hour, cards and SEPA debit
Choose GPUBank if
- H100 SXM at $1.85 per GPU-hour, against $3.63 to $3.79
- Per-second billing, where Scaleway's billing FAQ states a 60-minute minimum
- IPv4 included, plus interruptible and reserved discounts
- Pay in Bitcoin, USDT and 12+ other coins, from $25
Moving from Scaleway takes an afternoon.
Scaleway charges no egress on Instances, and its GPU OS images run Docker with the NVIDIA Container Toolkit, so containers move as they are. Object Storage is S3-compatible.
- 1Launch the same GPU on GPUBank in Paris or Frankfurt with a persistent disk.
- 2Copy block volume data with rsync, and Object Storage buckets with rclone.
- 3Run your Docker images or pick a template, then delete Scaleway volumes and flexible IPs.
# 1. a GPU and a disk on GPUBank gpubank launch --gpu h100-sxm --region eu-paris --disk data # 2. copy your data from the old machine rsync -azP old-host:/workspace/ root@<new-ip>:/workspace/ # 3. run the same image or template ssh root@<new-ip> 'cd /workspace && ./start.sh'
A few things you might be wondering.
Yes. Scaleway's H100-SXM-8-80G costs EUR 25.33 per hour, about $3.63 per GPU-hour converted and excluding VAT, and the 2-GPU size about $3.79. GPUBank charges $1.85 per GPU-hour for an H100 SXM, and $1.35 for an H100 PCIe against about $3.29 at Scaleway.
No. Scaleway accepts Visa, MasterCard, CB, UnionPay, American Express and SEPA debit, with a physical card required. GPUBank is paid from a USD balance topped up in Bitcoin, USDT, Ethereum and more than a dozen other coins.
Scaleway. It is a French company of the iliad Group with ISO/IEC 27001:2022 and HDS on all GPU Instances, a DPA and a 99.5 % SLA. GPUBank is also a French company with regions in Paris and Frankfurt, but publishes no SOC 2 or ISO 27001 report.
Scaleway bills GPU Instances per minute in EUR, postpaid monthly, with block storage and flexible IPv4 billed separately. Its billing FAQ still states a 60-minute minimum per resource. GPUBank bills per second from a prepaid USD balance, with IPv4, NVMe and traffic included.
Sources
Every figure about Scaleway was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01scaleway.com/en/pricing/gpu/GPU instance prices per zone (captured for PAR-1, PAR-2, WAW-2)
- 02scaleway.com/en/pricing/model-as-a-service/Managed inference GPU prices, Generative API free tier
- 03scaleway.com/en/gpu-instances/Per-minute billing, SLA 99.5%, HDS, no egress fees, Savings Plans exclusion
- 04scaleway.com/en/docs/gpu/reference-content/choosing-gpu-instance-type/GPU specs, NVLink, scratch storage, AI Supercomputer, SLO
- 05scaleway.com/en/docs/billing/faq/Payment methods, invoicing, minimum 60 minutes, suspension rules
- 06scaleway.com/en/docs/instances/faq/GPU per-minute billing, stopped instance billing, IPv4 price
- 07scaleway.com/en/pricing/network/Additional IPv4 price
- 08scaleway.com/en/pricing/storage/Block, object and file storage prices
- 09scaleway.com/en/product-availability-by-region/GPU availability by region
- 10scaleway.com/en/security-and-compliance/ISO 27001, HDS, SecNumCloud status, GDPR
- 11scaleway.com/en/support/Support tiers and prices
- 12scaleway.com/en/about-us/iliad Group subsidiary, countries of operation
- 13scaleway.com/en/terms/sla/SLA documents per product
- 14registry.terraform.io/providers/scaleway/scaleway/latest/docsOfficial Terraform provider
- 15scaleway.com/en/docs/instances/api-cli/API and CLI
What we could not verify (9)
- RENDER-S (P100) price not shown on the rendered pricing page; recorded in catalogue only.
- H100-SXM-1-80G and B300-SXM-1/2/4-288G appear in the quotas page but have no published price.
- Conflict between billing FAQ (minimum 60 minutes per resource) and Instances/GPU FAQ (GPU billed per minute); both kept.
- Flexible IPv4 price conflict: EUR 0.005/hour (network pricing) vs EUR 0.004/hour (Instances FAQ).
- Founding year not stated on the pages read.
- SOC 2 and HIPAA status not published.
- Serverless GPU (Serverless Containers with GPU) not verified; 'serverless' true refers to Generative APIs serverless and dedicated inference.
- Status page URL not verified by capture.
- SDK languages not verified on a captured page.
In short
GPUBank vs Scaleway, in facts.
The essentials of this comparison in one table, for quick readers and AI assistants. Prices last checked on 29 September 2026.
Read as Markdown- H100 SXM, per GPU-hour
- Scaleway: $3.63 to $3.79, EUR converted, excl. VAT
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Scaleway: $0.90, L4, excl. VAT
GPUBank: $0.23, RTX 4090 - Billing granularity
- Scaleway: Per minute
GPUBank: Per second - Stopped instance
- Scaleway: Compute stops; volumes, snapshots and IPs billed
GPUBank: Disk only - Persistent storage
- Scaleway: About $0.11 per GB-month (Block 5K)
GPUBank: $0.03 per GB-month - Egress traffic
- Scaleway: Included for Instances
GPUBank: Included - Crypto payments
- Scaleway: No
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Scaleway: Visa, MasterCard, CB, Amex, UnionPay; SEPA
GPUBank: No - Minimum deposit
- Scaleway: Not published
GPUBank: $25 - GPU models listed
- Scaleway: 6
GPUBank: 10 - Regions
- Scaleway: PAR-1, PAR-2, WAW-2
GPUBank: 4 - Serverless GPUs
- Scaleway: Serverless Generative APIs, managed inference
GPUBank: No - Availability target
- Scaleway: 99.5 % on GPU Instances
GPUBank: 99.9 % monthly - Compliance reports
- Scaleway: ISO/IEC 27001:2022, HDS; SOC 2 not published
GPUBank: None published - Support
- Scaleway: Free 24/7 tickets; paid plans from EUR 50/month
GPUBank: Tickets from a $25 balance - Headquarters
- Scaleway: Paris, France
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Scaleway was read on its own website, documentation or public API on 29 September 2026, with the address of each page listed above. A value we could not read on a primary source is marked "Not published" rather than estimated. GPUBank figures come from this site and its terms. Prices change: check both providers before you buy. Scaleway, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Scaleway.
