# OVHcloud GPU instances

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

## Pricing regimes (raw)

- **l40s-90, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **a10-45, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **a100-180, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **h100-380, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **h200-1920, on-demand (×8 node)** (resource): $0/vCPU-h, $0/GiB-h [alt]
- **l4-90, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **rtx5000-28, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h
- **t1-45, on-demand (×1)** (resource): $0/vCPU-h, $0/GiB-h

## 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, SSH, public IPv4, HTTPS preview URLs, egress allowlist, open internet, static egress IP, SOC 2, HIPAA, SSO, GPU, Windows, live fork (no pause), EU data residency, automatic snapshots, 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, browser, desktop GUI, computer-use API, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, self-hosting, BYOC, open source, arm64, macOS, Python, Node.js, HTTP method/path egress rules, wake on request, live resize, memory fork, MCP server, 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/ovh-gpu.json → feature_evidence

## Caveats

- International USD page, ex VAT. Bundled CPU/RAM/storage. Region selection affects rate/availability. IPv4 starts billing separately October 1, 2026; not yet applicable on research date. Page mislabels multiple GPU memory technologies: do not copy HBM2 claims for L4/H100.
- 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.
- Modes with gpu=null are intentionally excluded, not free GPU modes. Never substitute 0 for a null rate. Flags alt indicate DIY/ML infrastructure, not the same thing as managed untrusted-code sandboxes.
- Unknown fee, storage or bandwidth means full monthly total may be unavailable. Rates exclude taxes.
- t1-45 (Tesla V100) and rtx5000-28 (Quadro RTX 5000) are older-generation GPUs but still listed and orderable on the live price page (2026-09-28: t1-45 $0.77/h in BHS5/GRA5/GRA7/GRA9; rtx5000-28 $0.60/h in GRA9 only), so they are not flagged legacy. Their GPU names are outside the engine GPU ladder, so they are never used as an H100 substitute.

## How this provider charges

As of 2026-09-28. Public Cloud GPU instances.

International USD page, ex VAT. Bundled CPU/RAM/storage. Region selection affects rate/availability. IPv4 starts billing separately October 1, 2026; not yet applicable on research date. Page mislabels multiple GPU memory technologies: do not copy HBM2 claims for L4/H100.

| Regime | When applicable | Billing | Numbers ($/GPU-hour unless slice) | Source |
|---|---|---|---|---|
| on-demand  | l40s-90 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 1.8; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | a10-45 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 1; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | a100-180 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 3.07; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | h100-380 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 2.99; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | h200-1920 ; count [8]; International USD price page default; select region before purchase. | per-hour; min unknowns | 6.195; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | l4-90 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 1; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | rtx5000-28 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 0.6; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |
| on-demand  | t1-45 ; count [1]; International USD price page default; select region before purchase. | per-hour; min unknowns | 0.77; list; per_gpu_hour | [source](https://www.ovhcloud.com/en/public-cloud/prices/) |

## Gotchas
- International USD page, ex VAT. Bundled CPU/RAM/storage. Region selection affects rate/availability. IPv4 starts billing separately October 1, 2026; not yet applicable on research date. Page mislabels multiple GPU memory technologies: do not copy HBM2 claims for L4/H100.
- 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.

Explicit GPU sensitivity example: add **1 × L40S** to each of those 50 workers at the l40s-90 quoted shape. Compute component = 8,800 × $1.8 = **$15,840.0000**. CPU/RAM are included in that SKU (not independently resizable).
This is compute-only, before platform fees, storage, egress, setup, minimum rounding, cold-start/idle tails and capacity quotas. 50 concurrent GPUs are not promised by a unit rate.

## 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.