Nebius
Single-GPU to multi-node work on NVIDIA HGX H100, H200, B200 and B300, rented per GPU-hour with per-second billing and 1-GPU presets
L40S 48 GB · from:
$1.55≈ €1.38 (estimated conversion from $1.55)
L40S hosting for AI inference: prices per GPU-hour, EU and US regions
We may earn commissions when you shop through the links below. Learn more
The L40S is a 48 GB PCIe data-centre GPU often used for inference, fine-tuning and rendering. Compare verified L40S offers by price type, billing and region.
L40S hosting providers
L40S at a glance
Specifications come from the manufacturer's published data as recorded in our GPU catalog. Prices are USD per GPU-hour across the verified offers listed on this page.
| Figure | L40S 48 GB |
|---|---|
| Memory | 48 GB GDDR6 |
| FP16 tensor (dense) | 362.05 TFLOPS |
| Architecture · form factor | Ada Lovelace · PCIe |
| Providers listing it here | 5 |
| Lowest listed price | $0.749≈ €0.669 (estimated conversion from $0.749) per GPU-hour |
| Median pay-as-you-go price | $1.72≈ €1.54 (estimated conversion from $1.72) per GPU-hour |
| Specification source | NVIDIA L40S 48GB (opens in a new tab) |
Rent or buy an L40S?
Enter a purchase price per GPU from a quote you have; hardware prices are not in our data. We compare it with the median verified pay-as-you-go price on this page.
The L40S median on this page is $1.72 per GPU-hour. Enter a purchase price to see after how many rental hours buying would cost the same.
Providers without a verified L40S offer
These 5 providers have no verified L40S offer in our data (full provider review ): Fly.io, Hetzner, Hyperstack, Lambda and Together AI.
- Fly.io: “No GPUs: GPU Machines (A10, L40S, A100) became unavailable after 1 August 2026, ruling out self-hosted GPU inference, fine-tuning and training”
Tools for this decision
- GPU Cost CalculatorEstimate monthly GPU spend from verified hourly prices, usage hours, instances and storage.
- VRAM / Model FitEstimate the GPU memory a model needs for inference or fine-tuning, and which GPUs fit.
- Agent Hosting SelectorAnswer a few questions about your agent and get a best-fit hosting category and provider shortlist.
How we compare
Prices per GPU-hour.
GPU prices are USD per single GPU per hour from the provider's own pricing page or API. Multi-GPU nodes are divided by their GPU count, and a price only rented as a node says so (“rented as 8 GPUs”). Rows lead with the cheapest pay-as-you-go price (on-demand, secure, community or serverless). A spot, reserved or marketplace price leads only when the provider has no pay-as-you-go price here, or when you filter by that billing; it is labelled with its price type and listed in its own group after pay-as-you-go prices, never ranked against them. Prices exclude VAT and sales tax.Not verified means not verified.
When no official source confirms a value we show “Not verified” instead of an estimate. Where a provider's terms do not allow its prices to be republished, we link to its live pricing instead.Freshness on every row.
Updated today: checked on the current day (UTC). Fresh: up to 7 days. Stale: up to 30 days. Needs refresh: older than that.No paid placement.
Results only change order with the sort you pick. Affiliate relationships never affect order, filters or which providers appear.
Filters match a provider when one of its listed offers meets every GPU, VRAM, region, price and billing-increment condition together. Spot, reserved and serverless billing list each provider that offers that pricing and show only prices on that basis, or “not verified” when we have none. Read the full methodology and our affiliate disclosure.
Frequently asked questions
How does the L40S 48GB compare with the L40 48GB?
The L40S 48GB has 48 GB of GDDR6 at 864 GB/s and 362.05 dense FP16 TFLOPS. The L40 48GB has 48 GB of GDDR6 at 864 GB/s and 181.05 dense FP16 TFLOPS. Figures come from NVIDIA's published specifications as recorded in our GPU catalog.
What are the limitations of the L40S?
NVIDIA's L40S specifications list a dual-slot PCIe Gen4 card with 48 GB of memory and no NVLink or Multi-Instance GPU, so multi-GPU jobs communicate over PCIe and a model has to fit in 48 GB per card or be split across cards.
Are prices per GPU or per instance?
Per GPU. Every GPU price is shown in USD per single GPU per hour; offers for multi-GPU nodes are divided by the number of GPUs, and a price you can only rent as a whole node says so (for example “rented as 8 GPUs”). Storage, data transfer and taxes are not included unless the provider includes them.
Why do some values say “Not verified”?
We only show a value when an official provider source confirms it. If we could not confirm a price or feature, the row says “Not verified” instead of showing an estimate, and the provider page lists the sources we did check.