GPU MODELS / L40

L40 cloud GPU pricing

20 listings across 14 providers. Cheapest published rate: $0.190/GPU/hr at SpotGPUs. Sorted by per-GPU hourly price. Every row links to the provider's own published price.

WHAT IS THE L40

The L40 is the graphics-first sibling of the L40S: same 48 GB of GDDR6, full Ada RT cores, half the tensor throughput. It is a rendering, visualization and Omniverse card that also serves models competently. If the workload is AI-first, the L40S is the right variant.

MAKER
NVIDIA
ARCHITECTURE
Ada Lovelace
LAUNCHED
2022
MEMORY
48 GB GDDR6 ECC
MEMORY BANDWIDTH
864 GB/s
TENSOR COMPUTE
181 TFLOPS FP16 tensor dense (362 sparse)
POWER DRAW
300 W
INTERCONNECT
PCIe 4.0 x16

WHAT IT IS GOOD AT

Rendering and visualization
The design target - RT cores, 48 GB scene memory.
Light inference
Capable for models up to ~13B, but the L40S does it at similar cost.

WHO SHOULD NOT RENT THE L40

AI-first workloads - the L40S doubles tensor throughput for usually similar money.
70B-class anything.
ProviderConfigVRAMLocation$/hr$/GPU/hrSource
SpotGPUsSpotGPUsL40 × 148 GBNot published$0.190$0.190receipt
HydraHostHydraHostL40 × 148 GBNot published$0.550$0.550receipt
RunPodRunPodL40 × 148 GBGlobal$0.690$0.690receipt
AutoDLAutoDLL40 × 148 GBChina$0.716$0.716receipt
gpu.aigpu.aiL40 × 148 GBNot published$0.780$0.780receipt
Thunder ComputeThunder ComputeL40 × 148 GBNot published$0.790$0.790receipt
IO.NETIO.NETL40 × 148 GBNot published$0.790$0.790receipt
Massed ComputeMassed ComputeL40 × 148 GBNot published$0.860$0.860receipt
Hostnot GPUHostnot GPUL40 × 148 GBUS Central$0.860$0.860receipt
Hostnot GPUHostnot GPUL40 × 248 GBUS Central$1.720$0.860receipt
Massed ComputeMassed ComputeL40 × 448 GBNot published$3.440$0.860receipt
Hostnot GPUHostnot GPUL40 × 448 GBUS Central$3.440$0.860receipt
Hostnot GPUHostnot GPUL40 × 248 GBCanada Central$1.750$0.875receipt
Hostnot GPUHostnot GPUL40 × 148 GBCanada Central$0.880$0.880receipt
GPUBrazilGPUBrazilL40 × 145 GBAsia-Pacific$0.898$0.898receipt
HyperstackHyperstackL40 × 148 GBUS / EU / CA$1.000$1.000receipt
OblivusOblivusL40 × 148 GBNot published$1.050$1.050receipt
TensorDockTensorDockL40 × 148 GBMultiple regions$1.070$1.070receipt
GPUBrazilGPUBrazilL40 × 145 GBNorth America$1.194$1.194receipt
CoreWeaveCoreWeaveL40 × 848 GBNorth America / Europe$10.000$1.250receipt

THE L40 MARKET, BY THE NUMBERS

PROVIDERS
14
CONFIGURATIONS
20
MEDIAN $/GPU/HR
$0.86
SPREAD
6.6x

14 providers publish 20 L40 configurations on the panel today. Published per-GPU hourly rates run from $0.190 (SpotGPUs) to $1.25 (CoreWeave), a 7x spread between the cheapest and the most expensive published rate for the same chip. The median listing sits at $0.86/GPU/hr. 13 rows were reported in stock by the provider at last observation; the rest are published catalog rates. All rates are on-demand - committed-use and reserved pricing is excluded.

CHEAPEST L40 BY GEOGRAPHY
chinaAutoDL$0.716/hrunited statesHostnot GPU$0.860/hrcanadaHostnot GPU$0.875/hr
L40 PRICE TREND, DAILY PANEL SNAPSHOTS

Across every L40 row on the panel, the daily floor moved from $0.690 to $0.190/GPU/hr over 11 days of tracking (-72.5%). The median continuously-tracked listing held flat at $0.86. Snapshots run daily; the full history is public in the repo.

L40 PROVIDER NOTES
SpotGPUs · 1 config · 1 in stock
Sets the panel floor for this model.
from $0.190/hr
HydraHost · 1 config · 1 in stock
1 of 1 configs reported in stock at last check.
from $0.550/hr
RunPod · 1 config · 1 region · 1 in stock
1 of 1 configs reported in stock at last check.
from $0.690/hr
AutoDL · 1 config · 1 region
Published catalog rates.
from $0.716/hr
gpu.ai · 1 config · 1 in stock
1 of 1 configs reported in stock at last check.
from $0.780/hr
Thunder Compute · 1 config · 1 in stock
1 of 1 configs reported in stock at last check.
from $0.790/hr
IO.NET · 1 config
Published catalog rates.
from $0.790/hr
Massed Compute · 2 configs
Published catalog rates.
from $0.860/hr
BUYING L40: WHAT THE PANEL SAYS

At the current floor, one L40 running around the clock costs about $139 a month (730 hours of on-demand arithmetic). Node shape matters: the single-GPU floor is $0.190/GPU/hr against $1.250/GPU/hr on an 8-GPU node (+557.9%). Per-GPU floors by config size: 1x $0.19 · 2x $0.86 · 4x $0.86 · 8x $1.25. Compare the alternatives below before committing - the same budget often buys more than one chip class.

L40 VS THE ALTERNATIVES

L40 vs L40S
Same memory, double the tensor throughput on the S. AI workloads belong on the L40S; the L40 earns its place on graphics.

Compare live rates: L40 vs L40S

L40 PRICING FAQ

What is the cheapest L40 cloud GPU right now?
The cheapest published L40 rate on the panel is $0.190 per GPU-hour at SpotGPUs (L40 x 1, Not published). Every price links to the provider's own published rate.
How much does a L40 cost per hour?
Across 14 providers tracked today, published per-GPU hourly rates for L40 run from $0.19 to $1.25, with a median of $0.86. Committed-use and reserved rates are excluded; everything shown is on-demand.
How many providers rent L40 GPUs?
14 providers publish 20 distinct L40 configurations on Compute Cafe. 13 of those rows were reported in stock by the provider's live endpoint at last observation.
Where is L40 cheapest?
By published per-GPU hourly rate, the cheapest geography for L40 today is china at $0.72/hr (AutoDL). The spread across 3 tracked geographies runs to $0.88/hr in canada.
Are L40 prices going up or down?
Over the 11 days Compute Cafe has tracked the panel, the L40 floor is falling: $0.690 to $0.190/GPU/hr (-72.5%). The median continuously-tracked listing moved from $0.86 to $0.86 (+0.0%). Daily snapshots are public in the repo, so the trend is auditable.
What does a L40 cost per month?
At the current floor of $0.190/GPU/hr, a L40 running 24/7 costs about $139 per month (730 hours of on-demand arithmetic on the cheapest published rate). Committed-use contracts can price lower; the panel tracks on-demand rates only.
Is a full 8-GPU L40 node cheaper than a single card?
On today's panel: the single-GPU floor is $0.190/GPU/hr and the 8-GPU node floor is $1.250/GPU/hr - more expensive per GPU on a full node (+557.9%). Check the table for the exact configs behind each floor.
L40 or L40S?
Graphics, rendering, Omniverse: L40. Anything AI: L40S. They share memory and bandwidth; the difference is where NVIDIA spent the silicon.

See also alternatives to the L40 · cheapest GPU-hour per model · inference chips · training chips · by location