Scaleway
European sovereign cloud with GPU compute
Scaleway is under the market median on 1 of its 5 GPUs today.
Scaleway is a European cloud provider offering GPU instances, bare-metal servers, managed Kubernetes and Generative APIs, with a strong focus on EU data sovereignty and published data-center energy-efficiency metrics.
At a glance · prices collected Nov 21, 2026
What's good about Scaleway
- GPU lineup spanning L4 through B300-SXM (Blackwell) in 1-8 GPU configurations
- Two inference paths: serverless per-token Generative APIs and dedicated GPU deployments for the same model catalog
- European headquarters with data residency across multiple multi-AZ EU regions
- Free tier for trying Generative APIs, including a token allowance and audio transcription minutes
- 24/7 ticket support included on every plan, with published Power Usage Effectiveness and other sustainability metrics
Scaleway GPU prices
5 GPU models across 14 configurations. Rank compares Scaleway's cheapest on-demand listing for each GPU with every provider's cheapest.
| GPU | Scaleway | -60%median+100% | Rank | vs. median | Report |
|---|---|---|---|---|---|
| H100 SXM 8× | $3.48/hr | #33 of 56 providers | +12% | ||
| L40S 1× | $1.62/hr | #26 of 39 providers | +41% | ||
| HGX B300 8× | $8.24/hr | #13 of 34 providers | -13% | ||
| L4 1× | $0.87/hr | #17 of 21 providers | +24% | ||
| H100 PCIe 1× | $3.15/hr | #10 of 14 providers | +15% |
What a run costs on Scaleway
24 hours at today's cheapest listing for each GPU, at the size it is listed.
8× H100 SXM for 24 hours
At $3.48/GPU-hr
$668.12
$73.88 over the market median rate
1× L40S for 24 hours
At $1.62/GPU-hr
$38.77
$11.34 over the market median rate
8× HGX B300 for 24 hours
At $8.24/GPU-hr
$1,582.56
$239.52 under the market median rate
The market median is every provider's cheapest on-demand price for that GPU.
Scaleway LLM API prices
Pay-per-token pricing, per 1M tokens.
Prices last updated November 21, 2026
| Model | Input/1M | Output/1M | ||||
|---|---|---|---|---|---|---|
| $0.0033/min | - | |||||
| $0.165 | $0.659 | |||||
| $0.275 | $0.549 | |||||
| $0.275 | $1.65 | |||||
| $0.440 | $0.879 | |||||
| $0.659 | $3.96 | |||||
| $0.659 | $3.63 | |||||
| $0.659 | $1.98 | |||||
| $0.824 | $2.47 | |||||
| $0.989 | $0.989 | |||||
Scaleway pricing by model
Input, output, and batch rates, plus alternatives, for one model at a time.
Billing & terms
- On-demand GPU Instances
- Hourly per-instance billing for L4, L40S, H100 and B300-SXM configurations, with no reserved or spot options.
- Generative APIs - Serverless
- Pay-per-token billing for hosted chat, vision, embedding and audio models, with a discount for Batches API requests.
- Generative APIs - Dedicated Deployment
- Hourly billing by GPU resource type for isolated model hosting.
- Bare Metal
- Dedibox and Elastic Metal dedicated servers, billed separately from GPU compute.
- Regions
- GPU Instances span multiple multi-availability-zone regions in Europe; Generative APIs (serverless and dedicated) are currently served from the Paris region.
- Support
- 24/7 ticket support included on every plan; paid Advanced, Business and Enterprise tiers add hotline access, faster response times and, on Enterprise, a dedicated technical account manager. Additional resources include documentation, a community Slack and a public status page; support is offered in English and French.
Watch out for
- On-demand hourly pricing only; no reserved or spot pricing published for GPU Instances
- Some GPU types are limited to specific availability zones
- Faster support response times require a paid support plan
- Generative APIs dedicated deployment is billed hourly by GPU resource regardless of utilization
Features
- GPU Instances
- L4, L40S, H100 (PCIe and SXM) and B300-SXM (Blackwell) GPU instances in configurations from 1 to 8 GPUs.
- Generative APIs
- Serverless, pay-per-token access to hosted chat, vision, embedding and audio models, plus dedicated GPU deployments for models that need isolated capacity.
- Bare Metal
- Dedibox dedicated servers and cloud-integrated Elastic Metal alongside GPU and virtual instance compute.
- Managed Kubernetes
- Kapsule and Kosmos managed Kubernetes offerings for containerized workloads.
- EU Data Sovereignty
- Infrastructure across multiple multi-availability-zone European regions, with data residency in the EU.
- Automation Tooling
- Console, API, CLI and Terraform provider for provisioning and lifecycle automation.
Company
Published by Scaleway to identify the company behind this listing.
- Legal entity
- Scaleway SAS
- Headquarters
- 🇫🇷 Paris, France
- Registration
- RCS Paris 433 115 904
Getting started
Step 1
Create an account
Sign up in the Scaleway console and complete identity verification.
Step 2
Choose a region
Select an availability zone; some GPU types are limited to specific zones.
Step 3
Launch a GPU Instance or call a Generative API
Provision a GPU Instance from the console, API, CLI or Terraform, or call a hosted model through the Generative APIs.
Step 4
Connect and deploy
Connect to the running instance, or integrate the API endpoint into an application.
Step 5
Monitor usage and get support
Track consumption in the console and reach support via ticket, with paid plans offering hotline access.
Where this page's facts come from
- GPU prices
- Scaleway Product Catalog API (api.scaleway.com) · last collected Nov 21, 2026
- LLM API prices
- scaleway.com · last collected Nov 21, 2026
- Profile
- Reviewed Mar 14, 2026
- Method
- How we collect and check prices
Frequently Asked Questions
What GPU types does Scaleway offer?
Scaleway offers various GPU types including H100 SXM, H100 SXM, H100 SXM, HGX B300, L40S, L40S, L40S, L40S, H100 PCIe, H100 PCIe, L4, L4, L4, L4. Check the pricing table above for current availability and pricing.
How do I get started with Scaleway?
Create an account, Choose a region, Launch a GPU Instance or call a Generative API, Connect and deploy, Monitor usage and get support
What are Scaleway's main advantages?
Scaleway's main advantages include: GPU lineup spanning L4 through B300-SXM (Blackwell) in 1-8 GPU configurations, Two inference paths: serverless per-token Generative APIs and dedicated GPU deployments for the same model catalog, European headquarters with data residency across multiple multi-AZ EU regions, Free tier for trying Generative APIs, including a token allowance and audio transcription minutes, 24/7 ticket support included on every plan, with published Power Usage Effectiveness and other sustainability metrics.
What are Scaleway's limitations?
Scaleway's main limitations include: On-demand hourly pricing only; no reserved or spot pricing published for GPU Instances, Some GPU types are limited to specific availability zones, Faster support response times require a paid support plan, Generative APIs dedicated deployment is billed hourly by GPU resource regardless of utilization.
Compare Providers
Find the best prices for the same GPUs and models from other providers · All Scaleway alternatives, ranked →
All Scaleway Comparisons
The weekly price report
One email every Monday with the week's notable GPU and LLM API price moves, plus occasional product news. No spam.