PetaPrice

What does a PetaFLOP cost right now?

Live GPU rental prices from 50 providers, normalized per GPU-hour and ranked by what you actually get: dense FLOPS, VRAM or memory bandwidth per dollar.

Updated · 50/51 sources
Matching offers93872 GPUs · 46 providers · 405 live

Find the sweet spot

Cheapest offer for each GPU: price vs. dense FP16 / BF16 throughput. Up and to the left is better; the line connects GPUs nothing else beats on both.

NVIDIAAMDOtherSweet-spot frontierEqual $/PFLOP·hHollow: no stock signal (unverified), never ranked
$0.1$0.2$0.5$1$2$5$10$200.020.050.10.20.512Price per GPU-hour (log scale) →FP16 / BF16 PFLOPS per GPU →$1/PFLOP·h$3/PFLOP·h$10/PFLOP·h$30/PFLOP·hRTX 2080 TiRTX A4000RTX PRO 4000 BlackwellL40L40SRTX 6000 AdaGaudi 2RTX PRO 6000 BlackwellH100 SXMB300 SXM

Best value by GPU

Ranked by $ / PFLOP·h at each GPU's cheapest matching offer. Click a column header to re-sort, or a GPU to compare every provider.

Longer, greener bars mean more compute per dollar. The label shows how much more each GPU costs per unit than the best one. Click a GPU to compare every provider.

Data sources · 50/51 healthy
  • Vast.ai246 offers
  • RunPod67 offers
  • Verda92 offers
  • GPU.ai76 offers
  • Shadeform188 offers
  • Vultr4 offers
  • AWS pricing114 offers
  • Lambda28 offers
  • Nebius7 offers
  • Hyperstack30 offers
  • DigitalOcean30 offers
  • Together AI12 offers
  • Scaleway18 offers
  • OVHcloud40 offers
  • Cudo Compute0 offers
  • Hyperbolic10 offers
  • Oracle Cloud15 offers
  • Crusoe7 offers
  • Hot Aisle4 offers
  • TensorWave6 offers
  • Cirrascale5 offers
  • Seeweb44 offers
  • Azure Retail API74 offers
  • Massed Compute72 offers
  • Lium97 offers
  • Akash Network10 offers
  • CoreWeave34 offers
  • Oblivus46 offers
  • Thunder Compute4 offers
  • Koyeb0 offers
  • Akamai (Linode)13 offers
  • EmpirioLabs AI46 offers
  • Sesterce69 offers
  • Hetzner4 offers
  • STACKIT10 offers
  • T Cloud Public49 offers
  • IONOS4 offers
  • Gcorefailed
  • Exoscale22 offers
  • UpCloud76 offers
  • Leaseweb69 offers
  • Infomaniak42 offers
  • Leafcloud12 offers
  • Civo87 offers
  • Hivenet4 offers
  • Arkane Cloud2 offers
  • Latitude9 offers
  • SkyPilot: aws145 offers
  • SkyPilot: gcp76 offers
  • SkyPilot: nebius52 offers
  • SkyPilot: seeweb31 offers

Browse by provider

See which GPUs each cloud rents out and how its prices compare.

All providers →

Browse GPU prices

Every GPU we track right now, with its lowest on-demand price per hour.

GPU pricing questions

What is $ per PFLOP-hour?

It is the rental price of one GPU for one hour divided by its peak compute in PetaFLOPS at the precision you choose. It shows how much compute each dollar buys, so a cheap slow GPU and an expensive fast one can be compared directly. Lower is better.

Which GPU gives the most compute per dollar right now?

At dense FP16/BF16, the L40S currently leads at $1.05 per PFLOP-hour ($0.38 per GPU-hour at Lium). Rankings change with prices; the table above updates every 2 hours.

How much does an H100 cost per hour?

On-demand H100 SXM prices currently start at $1.73 per GPU-hour, with a median of $3.70 across 30 providers. Spot capacity starts at $0.92.

Why dense FLOPS instead of the numbers on NVIDIA's spec sheets?

Headline tensor numbers often assume 2:4 structured sparsity, which doubles them but is rarely used in practice. PetaPrice always uses dense throughput so GPUs and vendors are compared fairly.

Where does the price data come from?

Directly from the providers wherever they publish prices (public APIs, official price lists and pricing pages), with Shadeform's multi-cloud API and SkyPilot's catalogs filling gaps: 1389 offers from 50 providers, refreshed about every 2 hours. Prices are indicative list prices; confirm with the provider before booking.

What's the difference between on-demand, spot and reserved?

On-demand is billed by the hour with no commitment. Spot (interruptible) capacity is cheaper but can be reclaimed at any time. Reserved prices require a commitment, typically 1 to 3 years. PetaPrice never mixes them; the default view is on-demand.