Concierge · cluster-scale buying
Need 100+ GPUs?
Blocks at cluster scale price by conversation, not by rate card. We track verified capacity across every provider on this panel and quote the block directly.
Talk to us →
← ALL PROVIDERS
AWS

AWS

GPU pricing, tracked daily since 2026-09-22
LISTINGS
18
PRICE RANGE
$0.38 – $63.30/hr
GPU MODELS
14
REGIONS
US East (N. Virginia)
VISIT AWS →
PRICES BY GPU MODEL
Radeon Pro V5201 config
from $0.38/hr
T4G1 config
from $0.42/hr
T42 configs
from $0.53/hr
L42 configs
from $0.80/hr
A10G2 configs
from $1.01/hr
M601 config
from $1.14/hr
L40S2 configs
from $1.86/hr
V100 16GB1 config
from $3.06/hr
H100 SXM1 config
from $6.88/hr
Gaudi HL-2051 config
from $13.11/hr
A100 40GB1 config
from $21.96/hr
A100 80GB1 config
from $27.45/hr
V100 32GB1 config
from $31.21/hr
H2001 config
from $63.30/hr
ALL LISTINGS
Radeon Pro V520 8GBUS East (N. Virginia)
available$0.38/hr
T4G 16GBUS East (N. Virginia)
available$0.42/hr
T4 16GBUS East (N. Virginia)
available$0.53/hr
L4 24GBUS East (N. Virginia)
available$0.80/hr
A10G 24GBUS East (N. Virginia)
available$1.01/hr
M60 8GBUS East (N. Virginia)
available$1.14/hr
L40S 48GBUS East (N. Virginia)
available$1.86/hr
V100 16GBUS East (N. Virginia)
available$3.06/hr
4x T4 16GBUS East (N. Virginia)
available$3.91/hr
4x L4 24GBUS East (N. Virginia)
available$4.60/hr
4x A10G 24GBUS East (N. Virginia)
available$5.67/hr
H100 SXM 80GBUS East (N. Virginia)
available$6.88/hr
4x L40S 48GBUS East (N. Virginia)
available$10.49/hr
8x Gaudi HL-205 32GBUS East (N. Virginia)
available$13.11/hr
8x A100 40GBUS East (N. Virginia)
available$21.96/hr
8x A100 80GBUS East (N. Virginia)
available$27.45/hr
8x V100 32GBUS East (N. Virginia)
available$31.21/hr
8x H200 141GBUS East (N. Virginia)
available$63.30/hr
HOW AWS SITS IN THE MARKET

On chips where at least three providers publish rates (10 of AWS's 14 models), AWS is at or below the panel median on 3 and holds the single cheapest published rate on 0.

T4$0.53 vs $1.09 medianbelow median
A10G$1.01 vs $1.25 medianbelow median
V100 16GB$3.06 vs $3.06 medianbelow median
L4$0.80 vs $0.70 medianabove median
H100 SXM$6.88 vs $5.49 medianabove median
L40S$1.86 vs $1.39 medianabove median
H200$7.91 vs $4.59 medianabove median
A100 80GB$3.43 vs $1.65 medianabove median
A100 40GB$2.74 vs $1.19 medianabove median
V100 32GB$3.90 vs $0.42 medianabove median
AWS FAQ
What GPUs does AWS rent?
AWS publishes 18 configurations across 14 GPU models on the panel: Radeon Pro V520 from $0.38/hr, T4G from $0.42/hr, T4 from $0.53/hr, L4 from $0.80/hr and 10 more. Every rate links to AWS's own published source.
How much does AWS cost per hour?
Published hourly rates at AWS run from $0.38 to $63.30 across its tracked configurations. All figures are on-demand published rates - committed-use and reserved pricing is excluded from the panel.
Is AWS cheaper than other GPU clouds?
On the 10 GPU models where at least 3 providers publish rates, AWS is at or below the panel median on 3 and holds the cheapest published rate on 0. Compare row by row above - every price links to the provider's own page.
Where are AWS's GPUs located?
AWS's tracked configurations are in US East (N. Virginia).
Does Compute Cafe earn money from AWS links?
No. This provider has no affiliate arrangement with Compute Cafe; links go to its public pages.
SOURCE RECEIPT

On-demand hourly whole-instance USD read live from AWS's own pricing feed (verified 2026-09-22; the no-auth metered-unit map that powers the EC2 on-demand pricing page). GPU specs verified against AWS docs (CC-BY-SA). Exact linear multiples, duplicate 1-GPU configs, fractional-GPU sizes and Trainium/Inferentia excluded per panel policy. Rates, not stock - availability shows as catalog.

BILLING: per-second (partial instance-hours billed per-second for Linux) - verified 2026-09-23
AFFILIATE: none