Loading...
Side-by-side pricing across 27 marketplace listings. The L40S is 55.6% cheaper at median.
Side-by-side technical and market data. All prices in USD, listing-weighted median.
| Metric | H100 SXM | L40S |
|---|---|---|
| Median Price (USD/hr) | $3.29 | $1.46 |
| Marketplaces Tracking | 14 | 13 |
| VRAM | 80 GB | 48 GB |
| FP16 TFLOPS (dense) | 989 | 366 |
| Category | Datacenter GPUs | |
See individual pages: H100 SXM pricing · L40S pricing
Total rental cost at the listing-weighted median price for each GPU.
| Duration | H100 SXM | L40S | You Save |
|---|---|---|---|
| 1 hour | $3.29 | $1.46 | $1.829 |
| 1 day (24h) | $78.96 | $35.06 | $43.90 |
| 1 week (168h) | $552.72 | $245.45 | $307.27 |
| 1 month (730h) | $2,401.70 | $1,066.53 | $1,335.17 |
Over a typical month, choosing the L40S saves $1,335.17 compared to the H100 SXM. That's a 55.6% cost reduction.
Decision logic based on workload type, derived from VRAM and FP16 compute heuristics.
80GB VRAM vs 48GB — fits larger models.
300.61 TFLOPS/$ vs 250.51 — better cost efficiency.
Marketplace partners carrying these GPUs, with their pricing for each side.
| Marketplace | H100 SXM | L40S |
|---|---|---|
| io.net | $2.09 | $0.89 |
| Verda | $2.29 | $0.91 |
| Akash | $2.56 | $0.80 |
| Massed Compute | $2.75 | $0.88 |
| RunPod | $2.99 | $0.83 |
| Cyfuture AI | $3.38 | $1.27 |
| Vast.ai | $4.67 | $0.93 |
| Qubrid AI | $3.83 | $2.48 |
Other Datacenter GPUs GPU comparisons with similar price profiles.
Get price-drop alerts when either H100 SXM or L40S hits your target price at any marketplace.