Grab a GPU that is actually in stock

Price tables list capacity that does not exist. We index 874 configurations across 22 providers and 71 GPU models, and mark which ones a provider confirmed it can still give you. Right now 46 of 71 models are rentable — the rest are priced everywhere and available nowhere.

71 GPU models 22 providers 874 configurations 46 in stock now

Grab one now Open the price index

Prices and stock checked 1 minute ago

In-demand GPUs

Nvidia H100

$1.11/GPU/hr

16 providers · 8 in stock · 80 GB

Lium is cheapest, median $2.49

Nvidia A100

$0.27/GPU/hr

12 providers · 9 in stock · 80 GB

Vast.ai is cheapest, median $1.38

Nvidia H200

$0.50/GPU/hr

10 providers · 3 in stock · 141 GB

Runpod is cheapest, median $2.67

Nvidia L40S

$0.35/GPU/hr

10 providers · 6 in stock · 48 GB

Lium is cheapest, median $0.83

Nvidia RTX PRO 6000

$0.50/GPU/hr

9 providers · 4 in stock · 96 GB

Runpod is cheapest, median $1.35

Nvidia B200

$3.74/GPU/hr

7 providers · 3 in stock · 180 GB

BoostRun is cheapest, median $5.31

Nvidia RTX 6000 Ada

$0.45/GPU/hr

7 providers · 4 in stock · 48 GB

Vast.ai is cheapest, median $0.74

Nvidia RTX A6000

$0.32/GPU/hr

7 providers · 4 in stock · 48 GB

Lium is cheapest, median $0.57

All 71 GPU models →

Cheapest with confirmed stock

All 46 models in stock →

Popular comparisons

Browse by country

Frequently asked questions

What is the cheapest cloud GPU?

As of September 6, 2026 the cheapest GPU with confirmed stock is the Nvidia GTX 1060 at $0.03 per GPU hour on Vast.ai.

How much can I save by comparing providers?

The typical GPU model is 20% cheaper at its cheapest provider than at the median one. On a card that runs continuously, that gap is thousands of dollars a year for identical hardware.

Do you show availability, not just price?

Yes. Every row carries the stock the provider reported at our last check, and the in-stock page ranks only capacity that was confirmed. A price with no machine behind it is the most common way a cheap provider ends up costing more.