H200 SXM5 vs A30
NVIDIA H200 SXM5 (Hopper, 141 GB) against NVIDIA A30 (Ampere, 24 GB): memory, compute, power and rental price, compared for LLM inference and training.
Pick two GPUs to compare
Side-by-Side Specifications
| Spec | H200 SXM5 | A30 |
|---|---|---|
| Architecture | Hopper | Ampere |
| Memory | 141 GB HBM3e | 24 GB HBM2 |
| Memory bandwidth | 4,800 GB/s | 933 GB/s |
| FP16 tensor compute | 1,979 TFLOPS | 330 TFLOPS |
| INT8 tensor compute | 3,958 TOPS | 661 TOPS |
| Interconnect | NVLink 4.0 · 900 GB/s | NVLink 3.0 · 200 GB/s |
| TDP | 700 W | 165 W |
| Est. on-demand price | ~$10.00/h | ~$1.20/h |
| FP16 TFLOPS per $/h | 198 | 275 |
Highlighted values indicate the stronger spec. Hourly rates are indicative on-demand estimates.
Verdict
Raw performance: The H200 SXM5 leads on FP16 tensor compute (6.0x advantage), which translates directly into higher token throughput for inference and shorter training steps.
Memory: With 141 GB per card, the H200 SXM5 fits larger models on fewer GPUs — fewer cards means less inter-GPU communication and simpler deployments.
Value: At current on-demand rates, the A30 delivers more compute per dollar (275 vs 198 FP16 TFLOPS per $/h). If your model fits in its VRAM budget, it is usually the more economical choice.
GPUs Needed for Popular LLMs
Cards required to serve each model at 8-bit quantization (with 20% overhead for activations and KV cache).
| Model | VRAM (8-bit) | H200 SXM5 | A30 |
|---|---|---|---|
| GPT-5.6 Sol | 2682 GB | 20x | 112x |
| DeepSeek V4 Pro (671B) | 750 GB | 6x | 32x |
| Muse Spark 1.1 | 335 GB | 3x | 14x |
| Claude 5 Sonnet (175B) | 196 GB | 2x | 9x |
| Nova Premier (80B) | 89 GB | 1x | 4x |
| Nova Core (34B) | 38 GB | 1x | 2x |
| Nova Lite (12B) | 13 GB | 1x | 1x |
| Phi 3.5 (3.8B) | 4 GB | 1x | 1x |
Frequently Asked Questions
Which is better for LLM inference: H200 SXM5 or A30?
The H200 SXM5 delivers more raw FP16 compute (1,979 TFLOPS) and the H200 SXM5 offers the most memory per card (141 GB). For cost-efficiency, the A30 currently gives more FP16 TFLOPS per dollar of on-demand rental (275 vs 198 TFLOPS per $/h).
How much more memory does the H200 SXM5 have?
The H200 SXM5 has 141 GB of HBM3e versus 24 GB of HBM2 for the A30 — a ratio of 5.88x in favor of the H200 SXM5. More VRAM per card means fewer GPUs to fit a given model.
Is the H200 SXM5 or the A30 cheaper to rent?
Estimated on-demand rates are ~$10.00/h for the H200 SXM5 and ~$1.20/h for the A30. Raw hourly price is only part of the story: normalize by throughput (TFLOPS per $/h) and by how many cards you need for your model's VRAM.
How do the H200 SXM5 and A30 compare on power?
The H200 SXM5 has a TDP of 700W versus 165W for the A30. FP16 compute per watt: 2.8 vs 2.0 TFLOPS/W.
Deploy on a GPU cloud
Rent the H200 SXM5 or A30 by the hour instead of buying hardware.
