# Scaleway GPU instances

Official pricing: https://www.scaleway.com/en/pricing/gpu/  
Category: gpu-cloud · Isolation: vm

## Pricing regimes (raw)

- **B300-SXM-8-288G B300-SXM, on-demand (×8 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **H100-SXM-8-80G H100-SXM, on-demand (×8 node)** (resource): $0/vCPU-h, $0/GiB-h
- **H100-SXM-4-80G H100-SXM, on-demand (×4 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **H100-SXM-2-80G H100-SXM, on-demand (×2 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **H100-2-80G H100-PCIe, on-demand (×2 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **H100-1-80G H100-PCIe, on-demand (×1) (zones fr-par-2, pl-waw-2)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **L40S-8-48G L40S, on-demand (×8 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]
- **L40S-4-48G L40S, on-demand (×4 node)** (resource): $0/vCPU-h, $0/GiB-h [stock]

## Features

Yes: snapshots, fork/clone, pause/resume, persistent disk, volumes, ≥24 h sessions, custom image (Docker or snapshot), start from your own snapshot, full VM (own kernel), Docker inside, root, SSH, public IPv4, egress allowlist, static egress IP, private networking, SSO, GPU, Python, EU data residency, snapshots on demand, gVisor or VM (no shared kernel), inbound access rules, firewall inside (nftables), extra volumes, shared volumes

No: memory snapshots, idle auto-stop, your own Docker/OCI image, nested virtualization, browser, desktop GUI, computer-use API, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, HTTPS preview URLs, self-hosting, BYOC, open source, SOC 2, HIPAA, arm64, Windows, macOS, Node.js, HTTP method/path egress rules, wake on request, live resize, live fork (no pause), memory fork, MCP server, automatic snapshots, agent harness API, hosted agent API (their own agent), secret proxy, secret proxy for any API

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/scaleway-gpu.json → feature_evidence

## Caveats

- Embedded official catalog provides per-minute EUR ex-VAT rates and regional shapes. USD equivalents use ECB September 28, 2026 reference 1 EUR = 1.1378 USD; invoice is still EUR and bank FX/spreads may differ. native_node_hour=exact perMinute×60. Approximate monthly display is not a contractual cap.
- Any null count, CPU/RAM, billing increment or minimum is unverified, not unlimited/free.
- This is GPU-product research, not a claim that every feature of the entire vendor documentation was audited. Unverified features are null.
- Unknown fee, storage or bandwidth means full monthly total may be unavailable. Rates exclude taxes.
- Multi-GPU SKUs (2/4/8 GPUs) priced per GPU with gpu_min_count = N (sold only as N-GPU instances); the 'alt' flag was removed (ordinary GPU VMs, not an adjacent product). Per-GPU rates from the page __NEXT_DATA__ per-minute catalog (research/providers/scaleway-gpu.json, 2026-09-28); live page check: L4-2/4/8 EUR 1.58/3.15/6.30 per hour = 2/4/8 x L4-1 EUR 0.79.
- Block storage: Block Storage 5K EUR 0.000130/GB-h (~EUR 0.0949/GB-month, disk_gib_month converted at 1.1378) billed while the volume exists; 15K tier EUR 0.000177/GB-h; Block/Unified Snapshot EUR 0.000049/GB-h (scaleway.com/en/pricing/storage, 2026-09-28). GPU instances list "Block storage" (+ free scratch NVMe on L40S/H100/B300) with no included boot volume, so included_disk_gib 0.
- Egress: the GPU pricing page publishes no egress charge; Scaleway Virtual Instances say "List prices include egress and IPv6" but the GPU page does not repeat it, so egress_gib stays null (unverified, not free). Flexible IPv4 EUR 0.005/h (scaleway.com/en/pricing/network).
- Structure 2026-09-28: the catalog listed each SKU once per availability zone (fr-par-1/fr-par-2/pl-waw-2) at the same price; duplicates merged into one regime per SKU, label names the zones. Removed keys: on-demand-h100-2-80g-h100-pcie-5, on-demand-h100-1-80g-h100-pcie-7, on-demand-l40s-8-48g-l40s-9, on-demand-l40s-4-48g-l40s-11, on-demand-l40s-2-48g-l40s-13, on-demand-l40s-1-48g-l40s-15, on-demand-l4-8-24g-l4-17, on-demand-l4-8-24g-l4-18, on-demand-l4-4-24g-l4-20, on-demand-l4-4-24g-l4-21, on-demand-l4-2-24g-l4-23, on-demand-l4-2-24g-l4-24, on-demand-l4-1-24g-l4-26, on-demand-l4-1-24g-l4-27.
- Every GPU Instance needs a block boot volume (min 10 GB, 125 GB recommended for GPU OS): 125 GB x EUR 0.000130/GB-h = ~EUR 11.9 (~$13.5)/month while the volume exists, billed even when powered off. A flexible IPv4 (EUR 0.005/h) is created by default unless unticked.
- No commitment discount exists for GPU Instances: Compute Savings Plans (up to 25%) exclude B300, H100, RENDER, L4, L40S. Only on-demand pricing (or Elastic Metal monthly / sales contracts).
- Standby (stop-in-place) bills full compute; only 'power off' (archive) stops GPU billing. Kapsule GPU node pools are billed from creation, including GPU-operator setup time.
- Scaleway Billing FAQ says each Instance resource is "billed separately for a minimum of 60 minutes" per start/stop period, while the Instances FAQ says GPU Instances are "billed per minute of uptime". The 60-minute minimum is not confirmed for GPU Instances; if it applies, sessions under 1 h cost up to ~6x the per-minute estimate.

## How this provider charges

As of 2026-09-28. GPU Instances.

Embedded official catalog provides per-minute EUR ex-VAT rates and regional shapes. USD equivalents use ECB September 28, 2026 reference 1 EUR = 1.1378 USD; invoice is still EUR and bank FX/spreads may differ. native_node_hour=exact perMinute×60. Approximate monthly display is not a contractual cap.

| Regime | When applicable | Billing | Numbers ($/GPU-hour unless slice) | Source |
|---|---|---|---|---|
| on-demand  | B300-SXM-8-288G B300-SXM; count [8]; fr-par-2 | per-minute; min unknowns | USD 8.5335; EUR 7.5 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-SXM-8-80G H100-SXM; count [8]; fr-par-2 | per-minute; min unknowns | USD 3.60267303; EUR 3.16635 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-SXM-4-80G H100-SXM; count [4]; fr-par-2 | per-minute; min unknowns | USD 3.63271095; EUR 3.19275 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-SXM-2-80G H100-SXM; count [2]; fr-par-2 | per-minute; min unknowns | USD 3.76600422; EUR 3.3099 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-2-80G H100-PCIe; count [2]; fr-par-2 | per-minute; min unknowns | USD 3.2615037; EUR 2.8665 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-2-80G H100-PCIe; count [2]; pl-waw-2 | per-minute; min unknowns | USD 3.2615037; EUR 2.8665 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-1-80G H100-PCIe; count [1]; fr-par-2 | per-minute; min unknowns | USD 3.2615037; EUR 2.8665 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | H100-1-80G H100-PCIe; count [1]; pl-waw-2 | per-minute; min unknowns | USD 3.2615037; EUR 2.8665 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-8-48G L40S; count [8]; fr-par-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-8-48G L40S; count [8]; pl-waw-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-4-48G L40S; count [4]; fr-par-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-4-48G L40S; count [4]; pl-waw-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-2-48G L40S; count [2]; fr-par-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-2-48G L40S; count [2]; pl-waw-2 | per-minute; min unknowns | USD 1.672463598; EUR 1.46991 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-1-48G L40S; count [1]; fr-par-2 | per-minute; min unknowns | USD 1.672429464; EUR 1.46988 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L40S-1-48G L40S; count [1]; pl-waw-2 | per-minute; min unknowns | USD 1.672429464; EUR 1.46988 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-8-24G L4; count [8]; fr-par-1 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-8-24G L4; count [8]; fr-par-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-8-24G L4; count [8]; pl-waw-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-4-24G L4; count [4]; fr-par-1 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-4-24G L4; count [4]; fr-par-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-4-24G L4; count [4]; pl-waw-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-2-24G L4; count [2]; fr-par-1 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-2-24G L4; count [2]; fr-par-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-2-24G L4; count [2]; pl-waw-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-1-24G L4; count [1]; fr-par-1 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-1-24G L4; count [1]; fr-par-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |
| on-demand  | L4-1-24G L4; count [1]; pl-waw-2 | per-minute; min unknowns | USD 0.8960175; EUR 0.7875 (dated ECB FX); list-fx-converted; per_gpu_hour | [source](https://www.scaleway.com/en/pricing/gpu/) |

## Gotchas
- Embedded official catalog provides per-minute EUR ex-VAT rates and regional shapes. USD equivalents use ECB September 28, 2026 reference 1 EUR = 1.1378 USD; invoice is still EUR and bank FX/spreads may differ. native_node_hour=exact perMinute×60. Approximate monthly display is not a contractual cap.
- Any null count, CPU/RAM, billing increment or minimum is unverified, not unlimited/free.
- This is GPU-product research, not a claim that every feature of the entire vendor documentation was audited. Unverified features are null.
- GPU multi-count bundles, variant/region selection and fractional slices require a SKU-aware estimator; unsupported rows remain unpriced in strict cards.

## Worked example (required protocol workload)
4 vCPU / 8 GiB, 50 concurrent × 8 hours/day × 22 days = **8,800 instance-hours/month**. 30% CPU utilization does not reduce allocated GPU uptime. 50 GiB snapshots and 100 GiB egress are additional. The requested CPU-only workload has no GPU type/count, so its full GPU-provider total is **null**, not a fictitious CPU equivalent.

No fixed single-full-GPU USD quote suitable for a numeric example was verified. Compute = 8,800 × selected offer $/hour, or required node count × node price; keep the estimate null until the offer/FX/commitment is resolved.

## Evidence scope
Sources read are listed in the provider/features JSON. Raw pricing pages, relevant docs and API payloads are retained under `../raw/`. All unknown feature toggles remain null.