Hopper
NVIDIA H200 141GB
wins
4
The H200 upgrades the H100 with 141GB HBM3e memory and 4.8 TB/s bandwidth — a massive boost for memory-bound workloads like large LLM inference. Compute is identical to H100 at 989 TFLOPS FP16 (dense).
Ada Lovelace
NVIDIA RTX 4000 Ada 20GB
wins
1
The RTX 4000 Ada is a professional Ada Lovelace GPU with 20 GB GDDR6 and 53.4 TFLOPS FP16.
Performance comparison
Visual bars — winner highlighted
NVIDIA H200 141GB
Metric
NVIDIA RTX 4000 Ada 20GB
FP16 Tensor TFLOPS (dense)
Bandwidth
VRAM
TDP (lower=better)
TFLOPS/watt
GB/s per watt
Full specification table
| Spec | NVIDIA H200 141GB | NVIDIA RTX 4000 Ada 20GB |
|---|---|---|
| Architecture | Hopper | Ada Lovelace |
| VRAM | 141 GB HBM3e | 20 GB GDDR6 |
| VRAM type | HBM3e | GDDR6 |
| Memory bandwidth | 4,800 GB/s | 360 GB/s |
| FP16 Tensor TFLOPS (dense) | 989 TFLOPS | 53.4 TFLOPS |
| TDP | 700 W | 130 W |
| TFLOPS/watt | 1.41 T/W | 0.41 T/W |
| GB/s per watt | 6.86 GB/s/W | 2.77 GB/s/W |
| NVLink | Yes | No |
| Release year | 2024 | 2023 |
Live cloud pricing
On-demand pricing across providers — updated in real time
NVIDIA H200 141GB
NVIDIA RTX 4000 Ada 20GB
No live pricing data available.
Power efficiency analysis
TFLOPS/watt and GB/s/watt — critical for data center TCO
NVIDIA H200 141GB
NVIDIA RTX 4000 Ada 20GB
TFLOPS/watt measures compute efficiency — how much AI throughput you get per watt of power consumed. For data centers with PUE of 1.2–1.5, a 10% improvement in TFLOPS/watt translates directly to lower electricity costs and cooling requirements. GB/s/watt measures memory bandwidth efficiency, which is the binding constraint for memory-bound LLM inference workloads.
Performance-per-dollar
TFLOPS/$ and GB/s/$ at current best cloud prices
NVIDIA H200 141GB
NVIDIA RTX 4000 Ada 20GB
TFLOPS/$ = FP16 TFLOPS ÷ best on-demand $/hr. Higher is better — more compute per dollar. GB/s/$ = memory bandwidth ÷ best on-demand $/hr. For memory-bound inference workloads (most LLM serving), GB/s/$ is often a better predictor of cost-efficiency than TFLOPS/$. Prices are live from the Compute Comparison API and update in real time.
Workload Cost Comparison
Total job cost = best on-demand $/hr × GPU count × hours
| Workload | Scale | NVIDIA H200 141GB | NVIDIA RTX 4000 Ada 20GB |
|---|---|---|---|
| 7B LLM fine-tune | 8× GPU, 24 h | $501 | — |
| 70B LLM fine-tune | 64× GPU, 72 h | $12k | — |
| Inference server | 1× GPU, 730 h | $1.9k | — |
| Image gen batch | 4× GPU, 8 h | $84 | — |
Based on current best on-demand prices. Spot pricing typically 40–70% lower. Actual cost depends on provider, region, and availability.
When to choose each GPU
Choose NVIDIA H200 141GB for:
- Very large LLMs
- Memory-bound inference
- Multi-modal models
Weaknesses / Limitations
- Very limited cloud availability — fewer providers than H100
- Premium pricing over H100 with same FP16 TFLOPS
- 700W TDP — same power envelope as H100 with no efficiency gain
Choose NVIDIA RTX 4000 Ada 20GB for:
- Entry enterprise inference
- Low-power deployments
- Edge AI
Weaknesses / Limitations
- Only 20 GB VRAM — limits to ~13B models at FP16
- GDDR6 memory bandwidth far below HBM alternatives
- No NVLink — single-card VRAM ceiling
Popular GPU comparisons