Specifications Compared
| Spec | H200 SXM | L40S |
|---|---|---|
| TDP | 700W | 350W |
| VRAM | 141 GB | 48 GB |
| CUDA Cores | 16,896 | 18,176 |
| FP8 (dense) | 1,979 TFLOPS | 733 TFLOPS |
| Memory Type | HBM3e | GDDR6 |
| Architecture | Hopper | Ada Lovelace |
| FP16 (dense) | 989 TFLOPS | 362.05 TFLOPS |
| Form Factors | SXM | PCIe |
| INT8 (dense) | 1,979 TOPS | 733 TOPS |
| Interconnect | NVLink, PCIe 5.0, InfiniBand | PCIe 4.0 |
| Tensor Cores | 528 | 568 |
| FP32 Performance | 67 TFLOPS | 91.6 TFLOPS |
| FP64 Performance | 34 TFLOPS | 1.4 TFLOPS |
| Memory Bandwidth | 4,800 GB/s | 864 GB/s |
| FP8 (with sparsity) | 3,958 TFLOPS | 1,466 TFLOPS |
| FP16 (with sparsity) | 1,979 TFLOPS | 733 TFLOPS |
| INT8 (with sparsity) | 3,958 TOPS | 1,466 TOPS |
Performance Analysis
The FP16 dense figure reaches 989 TFLOPS on the H200 SXM family compared to 362.05 TFLOPS dense on the L40S family. This gap means the H200 SXM processes larger batches during training and inference when sparsity is unavailable. FP16 with sparsity reaches 1979 TFLOPS on the H200 SXM family versus 733 TFLOPS with sparsity on the L40S family so workloads that activate sparsity see proportional scaling. Memory bandwidth at 4800 GB/s on the H200 SXM family versus 864 GB/s on the L40S family directly limits batch sizes on the L40S during memory heavy operations. FP32 performance stands at 67 TFLOPS on the H200 SXM family against 91.6 TFLOPS on the L40S family so precision sensitive tasks without tensor cores favor the L40S in limited cases.
Current On-Demand Offers
Cheapest secure on-demand offer per provider that is reported in stock right now, per GPU per hour; multi-GPU instances show the whole-instance price alongside. Deploy opens the DeployGPU console for providers on the platform, otherwise the provider's own site. Stock is rechecked every minute.
H200 SXM
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| RunPod | global | 1 | $5.29 | — | Deploy |
| VERDA | FIN-HEL | 1 | $5.66 | — | Deploy |
| Lyceum | Europe | 1 | $6.19 | — | Deploy |
| LeaderGPU | The Netherlands | 1 | $6.22 | — | Deploy |
4 providers in stock, 4 offers. All H200 SXM offers, price history and alerts
L40S
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| Massed Compute | us-central-2 | 4 | $0.97 | $3.88 | Deploy |
| QuantaCloud | us-midwest-1 | 2 | $1.09 | $2.18 | Deploy |
| Ori | lille-4 | 1 | $1.55 | — | Deploy |
| Lyceum | Europe | 4 | $1.99 | $7.96 | Deploy |
4 providers in stock, 14 offers (cheapest per provider shown). All L40S offers, price history and alerts
Notify me when H200 SXM drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $5.29/GPU-hr.
QuantaCloud
Comparing H-series providers? We broker across all of them.
Hopper stock changes by the hour and prices differ by provider. If you need 16+ GPUs reserved or a cluster in the next 90 days, we quote H-series or B300 inventory at partner rates: one quote, 24h turnaround.
When to Choose the H200 SXM
The H200 SXM variant suits LLM training runs that require 141 GB HBM3e capacity to hold full models without partitioning. Its 4800 GB/s bandwidth supports sustained throughput on large context lengths where the L40S 48 GB limit forces smaller batches or model sharding.
When to Choose the L40S
The L40S variant fits inference pipelines that stay within 48 GB GDDR6 and operate at 350W TDP for dense deployments. Its 91.6 FP32 TFLOPS value aids scientific computing tasks that rely on non tensor operations without needing NVLink connectivity.
Use Cases
The H200 SXM 141 GB capacity holds entire large models without splitting across devices.
Higher dense FP16 throughput at 989 TFLOPS combined with 4800 GB/s bandwidth sustains larger batch inference.
The 141 GB HBM3e memory accommodates gradient and optimizer states for models exceeding 48 GB.
The L40S 48 GB GDDR6 suffices for standard image generation workloads at lower 350W power draw.
The L40S 91.6 FP32 TFLOPS value exceeds the H200 SXM 67 TFLOPS for non tensor workloads.
Frequently Asked Questions
What memory types distinguish the H200 SXM from the L40S?▾
The H200 SXM uses 141 GB HBM3e while the L40S uses 48 GB GDDR6. These capacities remain fixed across all variants of each family.
How does memory bandwidth differ between the two GPUs?▾
The H200 SXM family delivers 4800 GB/s bandwidth whereas the L40S family delivers 864 GB/s. Bandwidth figures apply uniformly to every variant within each family.
Which GPU offers higher dense FP16 performance?▾
The H200 SXM family reaches 989 TFLOPS dense FP16 compared to 362.05 TFLOPS dense FP16 on the L40S family. Both figures represent dense operation without sparsity.
What are the TDP values for the H200 SXM and L40S?▾
The H200 SXM family lists 700W TDP while the L40S family lists 350W TDP. These power figures apply to all variants of each respective family.
Does the H200 SXM support NVLink?▾
The H200 SXM variant includes NVLink interconnect support along with PCIe 5.0. The L40S variant supports only PCIe 4.0 without NVLink.
Which architecture powers each GPU?▾
The H200 SXM follows the Hopper architecture released in 2024. The L40S follows the Ada Lovelace architecture released in 2023.
Which is cheaper to rent, the H200 SXM or the L40S?▾
Cloud rental prices for both the H200 SXM and L40S vary by provider, configuration, and availability. This page shows live pricing from 25+ providers updated every 60 seconds. Scroll to the Live Cloud Pricing section to compare current rates.
How much VRAM does the H200 SXM have compared to the L40S?▾
The H200 SXM has 141 GB of HBM3e memory. The L40S has 48 GB of GDDR6 memory.
Can I find H200 SXM and L40S GPUs available to rent right now?▾
Yes. This page shows real-time availability across 25+ cloud GPU providers. The Live Cloud Pricing section displays only in-stock offers with current pricing.
What is the main difference between the H200 SXM and the L40S?▾
The H200 SXM uses the Hopper architecture (2024) while the L40S uses Ada Lovelace (2023). The H200 SXM delivers 2.7x the dense FP16 throughput (989 vs 362.05 TFLOPS, both without sparsity) and 5.6x the memory bandwidth of the L40S.
Rent these GPUs
Each GPU page lists every current on-demand offer by provider, per GPU-hour, updated every minute.
Related comparisons
How this page is made
- Specifications come from the NVIDIA, AMD and Intel datasheets for the H200 SXM and the L40S. Dense and sparse throughput are listed separately.
- Prices and availability are live: every offer shown is an in-stock, on-demand listing from a provider we track, rechecked every minute.
- The written comparison (overview, performance notes, when to choose each card, verdict, use cases and FAQ) was drafted with an AI model from the specification table above and passed an automated check that rejects any figure not in that table. It was last generated on . It contains no prices; those are always read live.
- Read how we collect the data, or report an error on this page.
Next steps
- Rent H200Every current offer by provider, daily price history and a price alert.
- Rent L40SEvery current offer by provider, daily price history and a price alert.
- GPU price indexHow on-demand prices for the major GPUs have moved, updated daily.
- All GPUsEvery GPU we track, with current lows and provider counts.