GPU rental pricing
V100 GPU pricing: rental cost per hour across cloud providers
Direct answer
The cheapest V100 GPU rental tracked by ByteCosts is about $0.17 per GPU-hour on-demand at Vast.ai (2x-Tesla_V100-32-65536 in North Carolina, US, NA), or roughly $0.34 per GPU-hour on spot. V100 is available from 7 of the providers in the catalog across 79 regions. Rates are cloud rental list prices per GPU-hour (instance price divided by GPU count), not card MSRP, and exclude committed-use discounts. Spot capacity is cheaper but interruptible, so size on-demand for steady jobs and use spot for checkpointed work.
Compare all GPUs across providers - Cheapest per-GPU-hour by accelerator →
V100 rental price by provider
Cheapest on-demand per-GPU-hour rate for V100 at each provider that offers it, from the ByteCosts catalog.
| Provider | Per-GPU/hr | Instance | GPUs | Region | Instance/hr |
|---|---|---|---|---|---|
| Vast.ai | $0.17 | 2x-Tesla_V100-32-65536 | 2 | North Carolina, US, NA | $0.35 |
| Lambda Labs | $0.79 | gpu_8x_v100_n | 8 | asia-northeast-1 | $6.32 |
| Paperspace | $2.29 | V100 | 1 | East Coast (NY2) | $2.29 |
| Google Cloud | $2.48 | 1 | asia-east1 | $2.48 | |
| Oracle Cloud | $2.95 | BM.GPU3.8 | 8 | af-johannesburg-1 | $23.60 |
| AWS | $3.06 | p3.16xlarge | 8 | us-east-1 | $24.48 |
| Azure | $3.06 | Standard_NC12s_v3 | 2 | eastus | $6.12 |
How to choose where to rent V100
The lowest per-GPU-hour rate is a starting point, not the decision. V100 cost in practice depends on utilization (an idle reserved GPU bills the full rate), whether your job tolerates spot interruption, and region and egress fees for moving weights or datasets.
Pick the cheapest provider that has capacity in a region near your data, can keep the GPU busy, and offers spot if your workload is checkpointed. A slightly higher headline rate in the right region often beats the cheapest row once egress and idle time are counted.
Frequently asked questions
How much does it cost to rent an V100 per hour?
The cheapest tracked V100 rental is about $0.17 per GPU-hour on-demand at Vast.ai, with other providers higher. See the table above for the rate at each provider. Prices are cloud rental list rates per GPU-hour, not card purchase price.
Which cloud is cheapest for V100?
In the ByteCosts catalog, Vast.ai has the lowest on-demand V100 rate at about $0.17 per GPU-hour, but availability and region matter as much as headline price. Compare the full per-provider table above and check live capacity before committing.
Is renting V100 cheaper than buying?
It depends on utilization. Renting wins for bursty or short-lived workloads and for avoiding capital outlay; owning can win only at sustained high utilization over a long horizon. ByteCosts tracks rental rates, not purchase price, so use these numbers for cloud cost planning.
Cite this page
V100 GPU rental pricing. ByteCosts. Updated June 15, 2026. https://bytecosts.com/gpu/v100/