L40S vs H200 rental price per hour

The H200 costs 215% more per hour than the L40S at the median and has 194% more memory (141 GB against 48 GB). Per GB of memory, the H200 is 7% dearer. Based on 12 L40S and 12 H200 public on-demand listings, last checked .

L40S median
$1.70
10 providers
H200 median
$5.35
12 providers
L40S memory
48 GB
Per GPU
H200 memory
141 GB
Per GPU

Side by side

L40S and H200 public on-demand prices

US dollars per GPU-hour from public pages. Every row behind these numbers is on the L40S price page and the H200 price page, each with its source link.

L40S and H200 price and memory side by side
MeasureL40SH200
Memory per GPU48 GB141 GB
Cheapest on-demand$0.97 (Massed Compute)$3.49 (Civo)
Median on-demand$1.70$5.35
Priciest on-demand$3.77$10.00
Providers with a price1012
Median price per GB of memory, per hour$0.035$0.038
What this page does not compare. Speed. Throughput depends on the model, precision and software, and we publish no benchmarks. This page compares public rental price and memory only.

Same provider

Providers that list both

Lowest public on-demand price per GPU-hour at each provider that lists both chips.

Providers that list both L40S and H200 on demand
ProviderL40S lowestH200 lowestH200 vs L40S
Amazon Web Services$1.86$7.91+325%
Civo$1.29$3.49+171%
CoreWeave$2.25$6.30+180%
Crusoe$1.50$4.29+186%
DigitalOcean$1.57$4.47+185%
Massed Compute$0.97$3.62+273%
Nebius$1.55$5.40+248%
Oracle Cloud (OCI)$3.50$10.00+186%
RunPod$1.09$5.29+385%
Verda$1.96$6.44+229%

FAQ

L40S vs H200 questions

Is the H200 more expensive than the L40S to rent?

At the median public on-demand price, yes by 215%: $5.35 against $1.70 per GPU-hour.

Which has more memory, the L40S or the H200?

The L40S has 48 GB and the H200 has 141 GB per GPU, as published by the vendors and shown in provider listings.

Where is the cheapest L40S and H200?

The cheapest public on-demand L40S we track is $0.97 at Massed Compute. For the H200 it is $3.49 at Civo. Last checked Nov 9, 2026.

Need L40S or H200 capacity?

Tell us the GPU, count, term, region and budget. We reply with the public prices that match and the checks to run before you pay. Free, and we do not sell compute.