GPU comparison

H100 SXM5 vs RTX 5090

Specifications and current cloud pricing, side by side.

HoppervsBlackwellUpdated 64 days ago

This variant comparison is part of the RTX 5090 vs H100 family comparison.

The H100 SXM5 wins for the most common use case of LLM training because its 989 TFLOPS dense FP16 performance and 80 GB memory capacity exceed the corresponding figures on the RTX 5090 variant by a wide margin.

H100 SXM5 from $2.50/GPU/hrRTX 5090 from $0.53/GPU/hr

Right now, from live stock

  • Cheapest right now: RTX 5090 at $0.53/hr on Vast.ai

    Deploy
  • Most providers in stock: H100 SXM5 (10)

    See all offers
  • Best $ per TFLOPS FP16 dense right now: H100 SXM5 ($0.0025 per TFLOPS-hour at 989 TFLOPS)

    Deploy

Specifications Compared

SpecH100 SXM5RTX 5090
TDP700W575W
VRAM80 GB32 GB
CUDA Cores16,89621,760
FP4 (dense)Not published1,676 TFLOPS
FP8 (dense)1,979 TFLOPS419 TFLOPS
Memory TypeHBM3GDDR7
ArchitectureHopperBlackwell
FP16 (dense)989 TFLOPS209.5 TFLOPS
Form FactorsSXM5PCIe
INT8 (dense)1,979 TOPS838 TOPS
InterconnectNVLink, PCIe 5.0, InfiniBandPCIe 5.0
Tensor Cores528680
FP32 Performance67 TFLOPS104.8 TFLOPS
FP64 Performance34 TFLOPS1.6 TFLOPS
Memory Bandwidth3,350 GB/s1,792 GB/s
FP4 (with sparsity)Not published3,352 TFLOPS
FP8 (with sparsity)3,958 TFLOPS838 TFLOPS
FP16 (with sparsity)1,979 TFLOPS419 TFLOPS
INT8 (with sparsity)3,958 TOPS1,676 TOPS

Performance Analysis

The H100 SXM5 delivers 989 TFLOPS dense FP16 performance compared to the 209.5 TFLOPS dense FP16 performance of the RTX 5090 while the RTX 5090 reaches 104.8 TFLOPS FP32 against the 67 TFLOPS FP32 of the H100 SXM5. This FP16 dense advantage on the H100 SXM5 supports faster matrix operations during training of large models and enables higher throughput in inference workloads that rely on dense tensor computations. The memory bandwidth of 3350 GB/s shared across H100 variants allows larger batch sizes than the 1792 GB/s bandwidth shared across RTX 5090 variants during both training and inference steps. When sparsity is applied the H100 SXM5 reaches 1979 TFLOPS FP16 with sparsity versus the 419 TFLOPS FP16 with sparsity on the RTX 5090 which further widens the gap for sparsity accelerated tasks.

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.

H100 SXM5

ProviderRegionGPUsPer GPU / hrInstance / hrDeploy
HyperstackCANADA-11$2.50—Deploy
QuantaCloudus-midwest-21$2.59—Deploy
Massed Computeus-central-31$2.73—Deploy
Oridallas-21$2.90—Deploy
LeaderGPUThe Netherlands8$3.59$28.75Deploy

10 providers in stock, 21 offers (cheapest per provider shown). All H100 SXM5 offers, price history and alerts

RTX 5090

ProviderRegionGPUsPer GPU / hrInstance / hrDeploy
Vast.aiSouth Korea, KR1$0.53—Deploy
LeaderGPUThe Netherlands1$2.29—Deploy

2 providers in stock, 13 offers (cheapest per provider shown). All RTX 5090 offers, price history and alerts

Which GPU to watchWatch the price of

Notify me when H100 SXM5 drops below a price

One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $2.50/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.

No waitlist24hr quote turnaroundInfiniBand fabric

Compare real-time pricing across 25+ providers

When to Choose the H100 SXM5

The H100 SXM5 serves as the better choice for workloads that require the 80 GB HBM3 memory capacity and NVLink interconnect to scale across multiple GPUs. Its 989 TFLOPS dense FP16 performance combined with 3350 GB/s bandwidth supports training of models that exceed the 32 GB limit of the RTX 5090 variant.

When to Choose the RTX 5090

The RTX 5090 serves as the better choice for workloads that benefit from its 104.8 TFLOPS FP32 performance and lower 575 W TDP in single GPU PCIe setups. Its 419 TFLOPS dense FP8 performance provides adequate throughput for inference tasks that fit within the 32 GB GDDR7 memory capacity.

Use Cases

LLM Training
H100 SXM5

The H100 SXM5 provides 989 TFLOPS dense FP16 performance and 80 GB memory capacity which exceeds the 32 GB on the RTX 5090.

LLM Inference
H100 SXM5

The H100 SXM5 delivers 1979 TFLOPS dense FP8 performance along with 3350 GB/s bandwidth that supports larger batch sizes than the RTX 5090.

Fine-tuning
H100 SXM5

The H100 SXM5 offers 80 GB HBM3 memory and NVLink interconnect for multi GPU scaling during fine tuning of large models.

Stable Diffusion
RTX 5090

The RTX 5090 supplies 104.8 TFLOPS FP32 performance and 575 W TDP suitable for single GPU image generation workloads.

Scientific Computing
H100 SXM5

The H100 SXM5 includes 989 TFLOPS dense FP16 performance and InfiniBand support for high performance computing clusters.

Frequently Asked Questions

What memory configurations distinguish the H100 SXM5 from the RTX 5090?▾

The H100 SXM5 variant includes 80 GB HBM3 memory while the RTX 5090 variant includes 32 GB GDDR7 memory. The H100 SXM5 also provides 3350 GB/s bandwidth compared to 1792 GB/s on the RTX 5090.

Which GPU offers higher dense FP16 throughput?▾

The H100 SXM5 reaches 989 TFLOPS dense FP16 while the RTX 5090 reaches 209.5 TFLOPS dense FP16. The H100 SXM5 also reaches 1979 TFLOPS dense FP8 compared to 419 TFLOPS dense FP8 on the RTX 5090.

How do the form factors and interconnects differ between these variants?▾

The H100 SXM5 uses an SXM5 form factor with NVLink and InfiniBand support while the RTX 5090 uses a PCIe form factor limited to PCIe 5.0 interconnect.

What TDP values apply to the H100 SXM5 and RTX 5090 variants?▾

The H100 SXM5 operates at a TDP of 700 W while the RTX 5090 operates at a TDP of 575 W.

Which is cheaper to rent, the H100 SXM5 or the RTX 5090?▾

Cloud rental prices for both the H100 SXM5 and RTX 5090 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 H100 SXM5 have compared to the RTX 5090?▾

The H100 SXM5 has 80 GB of HBM3 memory. The RTX 5090 has 32 GB of GDDR7 memory.

Can I find H100 SXM5 and RTX 5090 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 H100 SXM5 and the RTX 5090?▾

The H100 SXM5 uses the Hopper architecture (2022) while the RTX 5090 uses Blackwell (2025). The H100 SXM5 delivers 4.7x the dense FP16 throughput (989 vs 209.5 TFLOPS, both without sparsity) and 1.9x the memory bandwidth of the RTX 5090.

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 H100 SXM5 and the RTX 5090. 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