Specifications Compared
| Spec | A100 | H100 |
|---|---|---|
| TDP | 400W | 700W |
| VRAM | 40-80 GB | 80-94 GB |
| CUDA Cores | 6,912 | 16,896 |
| FP8 (dense) | Not published | 1,979 TFLOPS |
| Memory Type | HBM2e | HBM3 |
| Architecture | Ampere | Hopper |
| FP16 (dense) | 312 TFLOPS | 989 TFLOPS |
| Form Factors | SXM4, PCIe | SXM5, PCIe, NVL |
| INT8 (dense) | 624 TOPS | 1,979 TOPS |
| Interconnect | NVLink, PCIe 4.0, InfiniBand | NVLink, PCIe 5.0, InfiniBand |
| Tensor Cores | 432 | 528 |
| FP32 Performance | 19.5 TFLOPS | 67 TFLOPS |
| FP64 Performance | 9.7 TFLOPS | 34 TFLOPS |
| Memory Bandwidth | 2,039 GB/s | 3,350 GB/s |
| FP8 (with sparsity) | Not published | 3,958 TFLOPS |
| FP16 (with sparsity) | 624 TFLOPS | 1,979 TFLOPS |
| INT8 (with sparsity) | 1,248 TOPS | 3,958 TOPS |
Performance Analysis
FP16 dense performance reaches 312 TFLOPS on the A100 and 989 TFLOPS on the H100. The FP16 dense to FP32 ratio equals 312 divided by 19.5 on the A100 and 989 divided by 67 on the H100. This ratio indicates acceleration potential for mixed precision workloads in training and inference. Memory bandwidth of 3350 GB/s on the H100 compared with 2039 GB/s on the A100 supports larger batch sizes during model execution. FP16 with sparsity reaches 624 TFLOPS on the A100 and 1979 TFLOPS on the H100 so sparsity comparisons must stay separate from dense figures.
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.
A100
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| LeaderGPU | The Netherlands | 8 | $0.68 | $5.42 | Deploy |
| Vast.ai | Czechia, CZ | 1 | $1.05 | — | Deploy |
| ThunderCompute | USA | 1 | $1.09 | — | Deploy |
| Massed Compute | us-central-3 | 4 | $1.35 | $5.40 | Deploy |
| Hyperstack | CANADA-1 | 4 | $1.35 | $5.40 | Deploy |
9 providers in stock, 33 offers (cheapest per provider shown). All A100 offers, price history and alerts
H100
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| Hyperstack | CANADA-1 | 4 | $2.50 | $10.00 | Deploy |
| QuantaCloud | us-midwest-2 | 1 | $2.59 | — | Deploy |
| Massed Compute | us-central-3 | 1 | $2.73 | — | Deploy |
| RunPod | global | 1 | $2.89 | — | Deploy |
| Ori | dallas-2 | 2 | $2.90 | $5.80 | Deploy |
12 providers in stock, 24 offers (cheapest per provider shown). All H100 offers, price history and alerts
Notify me when A100 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $0.68/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 A100
The A100 suits environments that prioritize a TDP of 400W over higher throughput. Workloads that rely on INT8 dense performance of 624 TOPS without sparsity activation also favor the A100. Its 40 to 80 GB HBM2e configuration fits deployments where memory capacity beyond 80 GB is unnecessary.
When to Choose the H100
The H100 fits workloads that require FP16 dense performance of 989 TFLOPS or FP8 dense performance of 1979 TFLOPS. Its memory bandwidth of 3350 GB/s enables larger batch processing than the 2039 GB/s available on the A100. The 80 to 94 GB HBM3 capacity supports models that exceed 80 GB memory needs.
Use Cases
The H100 delivers FP16 dense performance of 989 TFLOPS against 312 TFLOPS on the A100.
The H100 provides FP8 dense performance of 1979 TFLOPS while the A100 does not publish an FP8 figure.
The H100 memory bandwidth of 3350 GB/s exceeds the A100 bandwidth of 2039 GB/s for efficient parameter updates.
Both GPUs support INT8 dense performance though the H100 reaches 1979 TOPS and the A100 reaches 624 TOPS.
The A100 TDP of 400W is lower than the H100 TDP of 700W for sustained power constrained runs.
Frequently Asked Questions
What is the memory bandwidth difference between A100 and H100?▾
The A100 lists memory bandwidth of 2039 GB/s. The H100 lists memory bandwidth of 3350 GB/s.
How does FP16 dense performance compare on A100 versus H100?▾
The A100 achieves 312 TFLOPS in FP16 dense. The H100 achieves 989 TFLOPS in FP16 dense.
What TDP values apply to the A100 and H100?▾
The A100 has a TDP of 400W. The H100 has a TDP of 700W.
Which GPU offers higher FP32 performance?▾
The A100 reaches 19.5 TFLOPS in FP32. The H100 reaches 67 TFLOPS in FP32.
Do the GPUs support the same interconnect options?▾
Both support NVLink and InfiniBand. The H100 adds PCIe 5.0 while the A100 uses PCIe 4.0.
Which is cheaper to rent, the A100 or the H100?▾
Cloud rental prices for both the A100 and H100 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 A100 have compared to the H100?▾
The A100 has 40 to 80 GB of HBM2e memory. The H100 has 80 to 94 GB of HBM3 memory.
Can I find A100 and H100 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 A100 and the H100?▾
The A100 uses the Ampere architecture (2020) while the H100 uses Hopper (2022). The H100 delivers 3.2x the dense FP16 throughput (989 vs 312 TFLOPS, both without sparsity) and 1.6x the memory bandwidth of the A100.
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 A100 and the H100. 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 A100Every current offer by provider, daily price history and a price alert.
- Rent H100Every 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.