L40S
The L40S is an NVIDIA Ada Lovelace data centre GPU with 48 GB of memory. It rents from $1.0900 per GPU-hour on-demand at RunPod, or $0.5000 if it may be interrupted. 34 offers from 8 providers.
Every offer
Price over time
- Cheapest
- Median
- Cheapest
- Median
The 8 machine types listed on 1 Mar are at $1.86 at best today, against $1.86 then. 13 more types have been listed since 22 Oct, and they set today's low of $1.57. 21 cloud machine types with this GPU, each in its cheapest region that day.
Price per GPU by box size
Per GPU, the cheapest 8-GPU box costs 1.5 times the cheapest single GPU.
Where it runs
What you would pay
8 L40S GPUs, 24 h a day, 30 days. Cheapest at RunPod
$6,278a month$13,882 less than the dearest
- RunPod$6,2788 × 1-GPU
- DigitalOcean$9,0438 × 1-GPU
- Scaleway$9,4871 × 8-GPU
- Vultr$9,6251 × 8-GPU
- OVHcloud$10,3684 × 2-GPU
- AWS$10,7198 × 1-GPU
- Oracle$20,1602 × 4-GPU
Each provider's cheapest way to cover 8 GPUs with the box sizes it sells; idle GPUs are paid for. List prices, before discounts and commitments.
All offers
Show the table of every offerHide the table
| Provider | Offer | Region | Price | GPUs | vCPU | Memory | Per GPU-hour | Box per hour | Checked |
|---|---|---|---|---|---|---|---|---|---|
| AWS | g6e.24xlarge | us-east-1 | Spot | 4 | 96 | 768 GiB | $0.5000 | $1.9999 | 34 min ago |
| Vast.ai | L40S:min | global | Marketplace, cheapest, 11 listings | 1 | — | — | $0.5363 | $0.5363 | 5 min ago |
| AWS | g6e.2xlarge | ap-south-2 | Spot | 1 | 8 | 64 GiB | $0.5643 | $0.5643 | 1 h ago |
| AWS | g6e.12xlarge | ap-south-2 | Spot | 4 | 48 | 384 GiB | $0.5952 | $2.3807 | 26 min ago |
| AWS | g6e.xlarge | ap-south-2 | Spot | 1 | 4 | 32 GiB | $0.7321 | $0.7321 | 26 min ago |
| RunPod | NVIDIA L40S | global | Community | 1 | — | — | $0.7900 | $0.7900 | 1 min ago |
| Vast.ai | L40S:median | global | Marketplace, median, 11 listings | 1 | — | — | $0.8004 | $0.8004 | 5 min ago |
| AWS | g6e.16xlarge | eu-south-2 | Spot | 1 | 64 | 512 GiB | $0.8424 | $0.8424 | 26 min ago |
| AWS | g6e.8xlarge | ap-south-2 | Spot | 1 | 32 | 256 GiB | $0.8603 | $0.8603 | 1 h ago |
| AWS | g6e.48xlarge | ap-south-2 | Spot | 8 | 192 | 1,536 GiB | $0.8640 | $6.9118 | 1 h ago |
| AWS | g6e.4xlarge | eu-central-1 | Spot | 1 | 16 | 128 GiB | $0.9367 | $0.9367 | 3 h ago |
| RunPod | NVIDIA L40S | global | On-demand | 1 | — | — | $1.0900 | $1.0900 | 1 min ago |
| Vultr | vbm-64c-2048gb-8-l40-gpu | global | Spot | 8 | 128 | 2,048 GiB | $1.4900 | $11.920 | 7 min ago |
| DigitalOcean | gpu-l40sx1-48gb | global | On-demand | 1 | 8 | 64 GiB | $1.5700 | $1.5700 | 7 min ago |
| Scaleway | L40S-8-48G | fr-par-2 | On-demand | 8 | 64 | 768 GiB | $1.6470 | $13.176 | 7 min ago |
| Scaleway | L40S-1-48G | fr-par-2 | On-demand | 1 | 8 | 96 GiB | $1.6470 | $1.6470 | 7 min ago |
| Scaleway | L40S-2-48G | fr-par-2 | On-demand | 2 | 16 | 192 GiB | $1.6470 | $3.2941 | 7 min ago |
| Scaleway | L40S-4-48G | pl-waw-2 | On-demand | 4 | 32 | 384 GiB | $1.6470 | $6.5882 | 7 min ago |
| Vultr | vbm-64c-2048gb-8-l40-gpu | global | On-demand | 8 | 128 | 2,048 GiB | $1.6710 | $13.368 | 7 min ago |
| Vultr | vcg-l40s-16c-180g-48vram | global | On-demand | 1 | 16 | 180 GiB | $1.6710 | $1.6710 | 7 min ago |
| Vultr | vcg-l40s-32c-375g-96vram | global | On-demand | 2 | 32 | 375 GiB | $1.6710 | $3.3420 | 7 min ago |
| Vultr | vcg-l40s-64c-750g-192vram | global | On-demand | 4 | 64 | 750 GiB | $1.6710 | $6.6840 | 7 min ago |
| OVHcloud | l40s-180 | global | On-demand | 2 | 30 | 180 GiB | $1.8000 | $3.6000 | 1 h ago |
| OVHcloud | l40s-360 | global | On-demand | 4 | 60 | 360 GiB | $1.8000 | $7.2000 | 1 h ago |
| OVHcloud | l40s-90 | global | On-demand | 1 | 15 | 90 GiB | $1.8000 | $1.8000 | 1 h ago |
| AWS | g6e.xlarge | us-east-1 | On-demand | 1 | 4 | 32 GiB | $1.8610 | $1.8610 | 1 h ago |
| AWS | g6e.2xlarge | us-west-2 | On-demand | 1 | 8 | 64 GiB | $2.2421 | $2.2421 | 1 h ago |
| AWS | g6e.12xlarge | us-east-1 | On-demand | 4 | 48 | 384 GiB | $2.6232 | $10.493 | 1 h ago |
| AWS | g6e.4xlarge | us-east-2 | On-demand | 1 | 16 | 128 GiB | $3.0042 | $3.0042 | 1 h ago |
| Oracle | BM.GPU.L40S.4 | global | On-demand | 4 | 224 | 1,024 GiB | $3.5000 | $14.000 | 1 h ago |
| AWS | g6e.24xlarge | us-east-2 | On-demand | 4 | 96 | 768 GiB | $3.7664 | $15.066 | 1 h ago |
| AWS | g6e.48xlarge | us-east-1 | On-demand | 8 | 192 | 1,536 GiB | $3.7664 | $30.131 | 1 h ago |
| AWS | g6e.8xlarge | us-east-1 | On-demand | 1 | 32 | 256 GiB | $4.5286 | $4.5286 | 1 h ago |
| AWS | g6e.16xlarge | us-east-2 | On-demand | 1 | 64 | 512 GiB | $7.5772 | $7.5772 | 1 h ago |
Get L40S prices from the API
curl "https://prices.cloudadapt.io/api/v1/gpus?model=L40S"
Per GPU-hour is the box price divided by its GPUs. Cloud machines appear once, in their cheapest region, except on the map. Spot, community and marketplace prices can change or disappear at any time. US dollars, before discounts and taxes.