NVIDIA

NVIDIA · Ada Lovelace · 2023

NVIDIA L40S.

Rent the NVIDIA L40S for AI training and inference. 48GB GDDR6, 183 TFLOPS FP16, deploy in 60 seconds.

48 GB
GDDR6
864 GB/s
Bandwidth
183
TFLOPS FP16
4× GPU
PCIe cluster
Available now · 4 regions

Cloud rental

$0.85/hr

On-demand · per-second billing

$0.78$0.92/hr range

Or $0.59/hr reserved (30% off)

Deploy

One command. NVIDIA L40S in 60 seconds.

Skip the dashboard if you don't need it. SDK, Python, or cURL — copy the snippet, paste your API key, ship.

import Runcrate from "@runcrate/sdk";

const rc = new Runcrate({ apiKey: "rc_live_••••••••••••••••" });

const instance = await rc.instances.create({
  gpu: "l40s",
  region: "auto",
  image: "runcrate/vllm:latest",
});

console.log(`SSH: ssh root@${instance.host}`);

Benchmarks

How NVIDIA L40S stacks up.

Side-by-side specs vs the closest alternatives. Bars are normalized to the highest value in each metric.

FP16

NVIDIA L40S183 TFLOPS
NVIDIA A4074.8 TFLOPS
NVIDIA A100 PCIe78 TFLOPS
NVIDIA RTX A600077.4 TFLOPS

VRAM

NVIDIA L40S48 GB
NVIDIA A4048 GB
NVIDIA A100 PCIe80 GB
NVIDIA RTX A600048 GB

Bandwidth

NVIDIA L40S864 GB/s
NVIDIA A40696 GB/s
NVIDIA A100 PCIe1.9 TB/s
NVIDIA RTX A6000768 GB/s

Full specs

NVIDIA L40S specifications.

Memory

VRAM48 GB
Memory typeGDDR6
Bandwidth864 GB/s

Compute

FP3291.6 TFLOPS
FP16183 TFLOPS
INT8733 TOPS
Tensor cores18,176
CUDA cores18,176

Power & form factor

TDP350 W
Form factorPCIe
ArchitectureAda Lovelace
Released2023

Cluster

Max GPUs/node4
InterconnectPCIe
ManufacturerNVIDIA
Available regions4

FAQ

Frequently asked.

Ready when you are

Deploy NVIDIA L40S
in 60 seconds.

Available across 4 regions · per-second billing · no commitments. Ship today.