Skip to prices

Cloud compute · per GPU-hour

GPU Price Index

A renter's tape of public cloud GPU prices. Not a broker, not a quote you can book here.

Print · 16 Oct 2026, 23:22 UTC

What a GPU costs to rent this hour.

15 providers returned a live price. Firm quotes are on-demand list rates, Runpod Community Cloud, and published starting prices. Spot and interruptible supply sits in its own column.

Firm H100
$1.99
H100 week
0.0%
Sources live
15
Models quoted
13
H100 $1.99H200 $3.00B200 $5.50B300 $6.94A100 80GB $0.802A100 40GB $0.456L40S $0.735L40 $0.495RTX 4090 $0.34RTX 5090 $0.497RTX A6000 $0.33RTX Pro 6000 $1.69GH200 $2.29

Cheapest firm quote

Single GPU when the provider sells one.

GPUProviderBoardType$/GPU-hrWeekSource
H100Runpod
H100 PCIe 80GB
Community$1.990.0%Page
H200Koyeb
H200 141GB
On-demand$3.00—Page
B200Koyeb
B200 180GB
On-demand$5.50—Page
B300Runpod
B300 288GB
Community$6.940.0%Page
A100 80GBVast.ai
Poland, PL
A100 SXM4 80GB
On-demand$0.802—API
A100 40GBVast.ai
California, US
A100 SXM4 40GB
On-demand$0.456—API
L40SVast.ai
Taiwan, TW
L40S 48GB
On-demand$0.735—API
L40Vast.ai
New York, US
L40 48GB
On-demand$0.495—API
RTX 4090Runpod
RTX 4090 24GB
Community$0.340.0%Page
RTX 5090Vast.ai
South Korea, KR
RTX 5090 32GB
On-demand$0.497+9.0%API
RTX A6000Runpod
RTX A6000 48GB
Community$0.330.0%Page
RTX Pro 6000Runpod
RTX Pro 6000 96GB
Community$1.690.0%Page
GH200Lambda
GH200 96GB
On-demand$2.29—Page

Biggest weekly moves

Compared with archived public pages and stored snapshots from about a week earlier. Down means cheaper for the renter.

  1. A40 · Runpod
    A40 48GB
    +20.4%
    $0.49 → $0.59
  2. L4 · Runpod
    L4 24GB
    +20.4%
    $0.49 → $0.59
  3. RTX 4090 · Runpod
    RTX 4090 24GB
    +20.3%
    $0.74 → $0.89
  4. RTX 5090 · Runpod
    RTX 5090 32GB
    +20.2%
    $0.99 → $1.19
  5. RTX Pro 6000 · Runpod
    RTX Pro 6000 96GB
    +19.1%
    $2.09 → $2.49
  6. H200 · Verda
    H200 SXM5 141GB
    +19.1%
    $5.07 → $6.04

Read this week in GPU prices

Cheapest firm price, by GPU

H100$1.99H200$3.00B200$5.50B300$6.94A100 80GB$0.80A100 40GB$0.46L40S$0.73L40$0.49RTX 4090$0.34RTX 5090$0.50RTX A6000$0.33RTX Pro 6000$1.69GH200$2.29
Dollars per GPU-hour. Each bar is the cheapest single-GPU firm quote in this print.

Flexible supply

Spot and interruptible rates, including Salad's priority tiers. These can disappear mid-job. They are not mixed into the firm column.

GPUProviderBoardType$/GPU-hrWeekSource
H100Hyperstack
H100 PCIe 80GB
Spot$2.000.0%Page
H200Verda
H200 SXM5 141GB
Spot$2.746+8.1%Page
B200Verda
B200 SXM6 180GB
Spot$3.955+7.8%Page
B300Verda
B300 SXM6 268GB
Spot$4.99+8.0%Page
A100 80GBVerda
A100 SXM4 80GB
Spot$1.008+7.8%Page
A100 40GBVerda
A100 SXM4 40GB
Spot$0.736+8.2%Page
L40SVerda
L40S 48GB
Spot$0.861+7.2%Page
L40Hyperstack
L40 48GB
Spot$0.800.0%Page
RTX 4090Salad
RTX 4090 · lowest 24GB
Interruptible$0.160.0%Page
RTX 5090Salad
RTX 5090 · lowest 32GB
Interruptible$0.250.0%Page
RTX A6000Verda
RTX A6000 48GB
Spot$0.368+7.3%Page
RTX Pro 6000Verda
RTX PRO 6000 96GB
Spot$1.163+8.7%Page