VERIFIED REAL-TIME SPOT RATE • 31 PROVIDERS INDEXED
NVIDIA H100 SXM5 (80GB HBM3) Spot Pricing
Continuously monitored across 31 GPU cloud platforms. The lowest verified spot rate is currently $1.99 / GPU-hr on LeaderGPU, delivering an immediate 56% cost reduction compared to AWS EC2 p5.48xlarge ($4.50/GPU-hr).
Cheapest Spot Instance
$1.99 / hr
Host: LeaderGPU • -56% vs AWS
Deploy Now on LeaderGPU →
Instant 1-Click Provisioning • Zero Waitlist
⚡ Spot Arbitrage Matrix: NVIDIA H100 SXM5 (80GB HBM3) vs AWS
| Metric | AWS On-Demand Baseline | Dynep Verified Spot (LeaderGPU) | Net Dollar Savings |
|---|---|---|---|
| Rate per GPU-hour | $4.50 / hr | $1.99 / hr | $2.51 / hr (56%) |
| 1x GPU Monthly Run (720 hrs) | $3,240 / mo | $1,433 / mo | +$1,807 / mo saved |
| 8x Node Monthly Run (720 hrs) | $25,920 / mo | $11,462 / mo | +$14,456 / mo ($173,472/yr) |
🔧 Hardware Specifications
- Accelerator:
- NVIDIA H100 SXM5 (80GB HBM3)
- VRAM Capacity:
- 80 GB
- Microarchitecture:
- Hopper 4nm
- FP16 Tensor TFLOPS:
- 1,979 TFLOPS
- Memory Bandwidth:
- 3.35 TB/s
🎯 Optimized Workloads
- ✓LLM Pre-training
- ✓70B Fine-tuning (LoRA / QLoRA)
- ✓Distributed High-Throughput Inference (vLLM)
Automate H100 Spot Provisioning
Query live H100 availability programmatically via CLI, Python SDK, or REST API.
Free Tier: 100 req/mo included
Developer CLI
# Query cheapest H100 spot rate in 1 second:
npx dynep-spot --gpu H100
cURL Live JSON Benchmark
curl -s "https://data.dynep.com/v1/spot/summary" | jq .
Claim Instant Developer API Key (No Card Required):
Dynamic GitHub README Badge
Embed real-time spot rates into your repository README. Displays live price and auto-refreshes every 5 minutes:
[](https://data.dynep.com/gpu/h100-lowest-price)