L40S · 48 GB · Ada Lovelace
L40S rental prices across providers
As of 12 November 2026, the lowest listed on-demand price for a L40S among 14 providers is $0.55 per GPU-hour at Runcrate. The median of each provider's lowest price is $1.54 and the highest is $3.50.
- Lowest listed
- $0.55
- Median
- $1.54
- Highest
- $3.50
- Providers
- 14
Vast.ai, per GPU-hour
of each provider’s lowest price
Oracle Cloud
33 configurations
By provider
L40S price by provider
- Vast.ai$0.47/GPU-hr
- Runcrateours$0.55/GPU-hr
- CloudRift$0.69/GPU-hr
- Beam$0.76/GPU-hr
- gpu.ai$0.87/GPU-hr
- Massed Compute$0.97/GPU-hr
- RunPod$1.09/GPU-hr
- Koyeb$1.20/GPU-hr
- Civo$1.29/GPU-hr
- Spheron$1.49/GPU-hr
- Crusoe$1.50/GPU-hr
- Nebius$1.55/GPU-hr
- DigitalOcean (Paperspace)$1.57/GPU-hr
- Vultr$1.67/GPU-hr
- OVHcloud$1.80/GPU-hr
- Amazon Web Services$1.86/GPU-hr
- Verda$2.00/GPU-hr
- CoreWeave$2.25/GPU-hr
- Oracle Cloud$3.50/GPU-hr
Every listing
All L40S configurations we track
| Configuration | Billing | Region | Source | ||||
|---|---|---|---|---|---|---|---|
| Vast.aimarketplace | 1× · Marketplace lowest listed (L40S) | not stated | $0.47 | $0.47 | not stated | 2026-11-12 | page |
| Runcrateours, live | 1× | per second | $0.55 | $0.55 | 5 regions | 2026-11-12 | page |
| CloudRift | 1× · On-demand (max 8 GPUs per instance) | per second | $0.69 | $0.69 | not stated | 2026-11-12 | page |
| Beammarketplace | 1× PCIe · On-demand machine (from price) | not stated | $0.76 | $0.76 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× · Marketplace median (L40S) | not stated | $0.80 | $0.80 | not stated | 2026-11-12 | page |
| gpu.aimarketplace | 1× · Community Cloud live floor (L40S) | per second | $0.87 | $0.87 | not stated | 2026-11-12 | page |
| gpu.aimarketplace | 1× · Non-community row live floor (L40S) | per second | $0.88 | $0.88 | not stated | 2026-11-12 | page |
| Massed Compute | 1× · L40S (48GB) | not stated | $0.97 | $0.97 | not stated | 2026-11-12 | page |
| Massed Compute | 2× · L40S (48GB) | not stated | $0.97 | $1.94 | not stated | 2026-11-12 | page |
| Massed Compute | 4× · L40S (48GB) | not stated | $0.97 | $3.88 | not stated | 2026-11-12 | page |
| Massed Compute | 8× · L40S (48GB) | not stated | $0.97 | $7.76 | not stated | 2026-11-12 | page |
| RunPod | 1× · Secure Cloud L40S | per second | $1.09 | $1.09 | not stated | 2026-11-12 | page |
| Koyeb | 1× · Serverless GPU instance L40S | per second | $1.20 | $1.20 | not stated | 2026-11-12 | page |
| Civo | 1× · Small 1 x NVIDIA L40S - 48GB | per hour | $1.29 | $1.29 | not stated | 2026-11-12 | page |
| Civo | 2× · Medium 2 x NVIDIA L40S - 48GB | per hour | $1.29 | $2.58 | not stated | 2026-11-12 | page |
| Civo | 4× · Large 4 x NVIDIA L40S - 48GB | per hour | $1.29 | $5.16 | not stated | 2026-11-12 | page |
| Civo | 8× · Extra Large 8 x NVIDIA L40S - 48GB | per hour | $1.29 | $10.32 | not stated | 2026-11-12 | page |
| Spheronmarketplace | 1× · Marketplace lowest-priced live on-demand (L40S) | per minute | $1.49 | $1.49 | not stated | 2026-11-12 | page |
| Crusoe | 1× · NVIDIA L40S 48GB | per hour | $1.50 | $1.50 | not stated | 2026-11-12 | page |
| Nebiusmarketplace | 1× · NVIDIA L40S with Intel CPU | not stated | $1.55 | $1.55 | not stated | 2026-11-12 | page |
| DigitalOcean (Paperspace) | 1× · NVIDIA L40S on-demand | per second | $1.57 | $1.57 | not stated | 2026-11-12 | page |
| Vultr | 1× · NVIDIA L40S (1 GPU) | per hour | $1.67 | $1.67 | not stated | 2026-11-12 | page |
| Vultr | 2× · NVIDIA L40S (2 GPU) | per hour | $1.67 | $3.34 | not stated | 2026-11-12 | page |
| Vultr | 4× · NVIDIA L40S (4 GPU) | per hour | $1.67 | $6.68 | not stated | 2026-11-12 | page |
| Vultr | 8× · NVIDIA L40S (8 GPU) | per hour | $1.67 | $13.37 | not stated | 2026-11-12 | page |
| OVHcloud | 1× · l40s-90 | per hour | $1.80 | $1.80 | not stated | 2026-11-12 | page |
| OVHcloud | 2× · l40s-180 | per hour | $1.80 | $3.60 | not stated | 2026-11-12 | page |
| OVHcloud | 4× · l40s-360 | per hour | $1.80 | $7.20 | not stated | 2026-11-12 | page |
| Nebiusmarketplace | 1× · NVIDIA L40S with AMD CPU | not stated | $1.82 | $1.82 | not stated | 2026-11-12 | page |
| Amazon Web Services | 1× · g6e.xlarge | per second | $1.86 | $1.86 | us-east-1 | 2026-11-12 | page |
| Verda | 1× · 1x L40S 48GB (20 vCPU, 60 GB RAM) | not stated | $2.00 | $2.00 | not stated | 2026-11-12 | page |
| Amazon Web Services | 1× · g6e.2xlarge | per second | $2.24 | $2.24 | us-east-1 | 2026-11-12 | page |
| CoreWeave | 8× · NVIDIA L40S | not stated | $2.25 | $18.00 | North America | 2026-11-12 | page |
| CoreWeave | 8× · NVIDIA L40S | not stated | $2.25 | $18.00 | Europe | 2026-11-12 | page |
| Amazon Web Services | 4× · g6e.12xlarge | per second | $2.62 | $10.49 | us-east-1 | 2026-11-12 | page |
| Amazon Web Services | 1× · g6e.4xlarge | per second | $3.00 | $3.00 | us-east-1 | 2026-11-12 | page |
| Oracle Cloud | 1× · Compute - GPU - L40S (B109479) | not stated | $3.50 | $3.50 | not stated | 2026-11-12 | page |
| Amazon Web Services | 4× · g6e.24xlarge | per second | $3.77 | $15.07 | us-east-1 | 2026-11-12 | page |
| Amazon Web Services | 8× · g6e.48xlarge | per second | $3.77 | $30.13 | us-east-1 | 2026-11-12 | page |
| Amazon Web Services | 1× · g6e.8xlarge | per second | $4.53 | $4.53 | us-east-1 | 2026-11-12 | page |
| Amazon Web Services | 1× · g6e.16xlarge | per second | $7.58 | $7.58 | us-east-1 | 2026-11-12 | page |
Questions
L40S rental: common questions
What is the lowest L40S price per hour?
As of 12 November 2026, the lowest listed on-demand price for a L40S among 14 providers is $0.55 per GPU-hour at Runcrate. The median of each provider's lowest price is $1.54 and the highest is $3.50.
How much does an 8-GPU L40S server cost per hour?
The lowest 8-GPU on-demand configuration in our data is Massed Compute at $0.97 per GPU, $7.76 per hour for the server (read 2026-11-12).
Which providers bill L40S time by the second?
Runcrate, CloudRift, gpu.ai, RunPod, Koyeb, DigitalOcean (Paperspace), and Amazon Web Services state per-second billing for their lowest-priced L40S listing.
Is the price per GPU or per server?
Per GPU per hour. The total column multiplies by the number of GPUs in that configuration. Multi-GPU configurations often cost a little more per GPU.
How current is this data?
Prices are read from each provider's public pricing page; the table shows the date each was read, and the latest read is 12 November 2026. Runcrate's own price comes live from runcrate.ai.
Billing as stated on each provider's page, per second: Runcrate, CloudRift, RunPod, Koyeb, DigitalOcean (Paperspace), and Amazon Web Services; per hour: Civo, Crusoe, Vultr, and OVHcloud. Providers that do not state it are not listed here.
Disclosure
Provider Index is published by the Runcrate team, who also run Runcrate, an AI cloud. Runcrate is listed in the table for L40S, marked "ours, live". It is sorted by price like every other row, with no pinned position. Other providers' prices are read from their own public pricing pages. We have no paid placement.