Two Nvidia data center GPUs, one Hopper and one Ada Lovelace: 80 GB HBM3 against 48 GB GDDR6, 3.9x the memory bandwidth on the H100, and NVLink 900 GB/s on the H100 where the L40S is PCIe only.
At a glance
Memory
- H100
- 80 GB
- L40S
- 48 GB
1.7x more on H100
Memory bandwidth
- H100
- 3,350 GB/s
- L40S
- 864 GB/s
3.9x higher on H100
Cheapest on-demand
- H100
- $1.41 /GPU/hr
- L40S
- $0.38 /GPU/hr
3.7x more expensive on H100
Price comparison
Median price per GPU per hour by billing type, with the cheapest listing under it.
| Billing | H100 | L40S | Difference |
|---|---|---|---|
| On-demand | $3.34 /GPU/hr from $1.41 | $1.40 /GPU/hr from $0.38 | 2.4x more expensive on H100 |
| Reserved | $3.11 /GPU/hr from $1.53 (1 mo) | $1.04 /GPU/hr from $0.49 (3 mo) | 3x more expensive on H100 |
| Spot | $2.00 /GPU/hr from $0.35 | $0.98 /GPU/hr from $0.20 | 2x more expensive on H100 |
Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Medians are across providers, one vote each, so a large catalog does not outweigh a small one. Verify before provisioning. More on how we price.
Price history
H100 vs L40S price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | H100 | L40S | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 80 GB HBM3 94 GB on NVL | 48 GB GDDR6 | 1.7x more on H100 |
| Bandwidth | 3,350 GB/s 2,000 GB/s on PCIe 3,938 GB/s on NVL | 864 GB/s | 3.9x higher on H100 |
| Bus width | 5,120-bit | 384-bit | 13x wider on H100 |
| Compute · dense | |||
| FP8 | 1,979 TFLOPS | 733 TFLOPS | 2.7x higher on H100 |
| FP16 / BF16 | 989.5 TFLOPS | 362.1 TFLOPS | 2.7x higher on H100 |
| INT8 | 1,979 TOPS | 733 TOPS | 2.7x higher on H100 |
| FP32 | 67 TFLOPS | 91.6 TFLOPS | 1.4x higher on L40S |
| FP64 | 34 TFLOPS | Not supported | H100 only |
| Platform | |||
| Interconnect | NVLink 900 GB/s NVLink 600 GB/s on PCIe, NVL | PCIe only | NVLink on H100 only |
| Board power | 700 W | 350 W | 2x higher on H100 |
| Released | Q3 2022 | Q3 2023 | 11 months newer on L40S |
| Precision | |||
| Supported formats | FP8FP16BF16TF32FP32FP64INT8 | FP8FP16BF16TF32FP32INT4INT8 | |
Figures are for each model's headline configuration. Where a form factor carries a different number, it is named under the figure.
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: H100 datasheet and L40S datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | H100 /GPU/hr | L40S /GPU/hr |
|---|---|---|
Lium
|
On-Demand $1.41 | On-Demand $0.38 |
Vast.ai
|
On-Demand $1.73 | On-Demand $0.53 |
GPU.ai
|
On-Demand $1.98 | On-Demand $0.79 |
Runpod
|
On-Demand $2.59 | On-Demand $0.79 |
HyperAI
|
On-Demand $1.80 | -- |
Koyeb
|
On-Demand $2.50 | -- |
Spheron
|
On-Demand $2.64 | -- |
Massed Compute
|
On-Demand $2.73 | -- |
Novita
|
-- | On-Demand $0.55 |
EmpirioLabs AI
|
-- | On-Demand $0.80 |
Packet·ai
|
-- | On-Demand $0.92 |
Sesterce
|
-- | On-Demand $0.97 |
53 providers list the H100 and 36 the L40S. Every listing is on the H100 page and the L40S page.
Similar GPUs
Other comparisons in the same class.
-
Nvidia B300
From $6.50 /GPU/hr Compare -
Nvidia H200
From $2.09 /GPU/hr Compare -
Nvidia RTX 5090
From $0.35 /GPU/hr Compare -
Nvidia A100
From $0.40 /GPU/hr Compare -
Nvidia L40
From $0.46 /GPU/hr Compare
Our data for Nvidia H100 was last updated on Sept. 14, 2026, and for Nvidia L40S on Sept. 14, 2026.