GPU Marketplace

Compare GPU cloud providers for training, fine-tuning, and inference.

Filters
✦ Free Tier Only: Pros & Cons
Pros: No credit card required on most plans · Great for prototyping and learning · Zero cost to get started.
Cons: Limited resources (CPU, RAM, bandwidth) · Often includes usage caps or sleep/spin-up delays · Not suitable for production traffic · May require upgrade without warning.
No providers match your filters.
AWS logo
AWS GPU Cloud
EC2 GPU Instances · Enterprise GPU compute for training and inference.
from $0.81/hour
US · EU · Asia · Australia · South America · Middle East · Africa

g6 (L4 24GB x1)
L4 · 1× · 24GB VRAM · inference
Best value for inference
$0.81
/hour
Spot: $0.10/hr
g5 (A10G 24GB x1)
A10G · 1× · 24GB VRAM · inference
Inference + light training
$1.01
/hour
Spot: $0.50/hr
g5 (A10G 24GB x4)
A10G · 4× · 24GB VRAM · inference
Multi-GPU inference
$3.58
/hour
Spot: $2.47/hr
p4d (A100 40GB x8)
A100 · 8× · 40GB VRAM · training
High-throughput training
$32.77
/hour
Spot: $13.07/hr
p5 (H100 80GB x8)
H100 · 8× · 80GB VRAM · training
Frontier training (H100)
varies
/hour
Spot: $57.76/hr
Google Cloud logo
Google Cloud GPU Cloud
Compute Engine GPUs · A2/A3 GPU families for training and inference.
from $0.71/hr
US · EU · Asia · Australia · South America

L4 24GB x1
L4 · 1× · 24GB VRAM · inference
$0.71
/hr
L4 24GB x4
L4 · 4× · 24GB VRAM · inference
$4.00
/hr
A100 40GB x1
A100 · 1× · 40GB VRAM · training
$3.67
/hr
A100 80GB x1
A100 · 1× · 80GB VRAM · training
$5.07
/hr
H100 80GB x8
H100 · 8× · 80GB VRAM · training
$88.49
/hr
Microsoft Azure logo
Microsoft Azure GPU Cloud
NV/NC/ND GPU VMs · Enterprise GPU instances across inference and training tiers.
from $0.66/hour
US · EU · Asia · Australia · South America · Middle East · Africa

NC4as T4 v3 (T4 16GB x1)
T4 · 1× · 16GB VRAM · inference
$0.66
/hour
ND96asr v4 (A100 80GB x8)
A100 · 8× · 80GB VRAM · training
$27.20
/hour
ND96isr H100 v5 (H100 80GB x8)
H100 · 1× · 80GB VRAM · training
$9.98
/hour
Lambda Labs logo
Lambda Labs GPU Cloud
Lambda GPU Cloud · Popular for cost-effective training and fine-tuning.
from $0.75/hour
US

A10 24GB x1
A10 · 1× · 24GB VRAM · inference
Budget inference / light workloads
$0.75
/hour
A100 40GB x1
A100 · 1× · 40GB VRAM · training
Training and fine-tuning
$1.29
/hour
A100 80GB x1
A100 · 1× · 80GB VRAM · training
More VRAM for larger jobs
$1.99
/hour
A100 80GB x8
A100 · 8× · 80GB VRAM · training
Multi-GPU training cluster
$14.32
/hour
H100 80GB x1
H100 · 1× · 80GB VRAM · training
Top-end training (availability varies)
$2.99
/hour
H100 80GB x8
H100 · 8× · 80GB VRAM · training
Full H100 pod
$23.92
/hour
RunPod logo
RunPod GPU Cloud
GPU Pods · On-demand and spot GPU pods.
from $0.74/hour
US · EU

RTX 4090 24GB x1
RTX 4090 · 1× · 24GB VRAM · inference
Great perf/$ for many workloads
$0.74
/hour
Spot: $0.44/hr
L40S 48GB x1
L40S · 1× · 48GB VRAM · inference
Strong inference + medium training
$0.79
/hour
Spot: $0.54/hr
L40S 48GB x4
L40S · 4× · 48GB VRAM · inference
Multi-GPU inference/training
$3.16
/hour
Spot: $2.16/hr
A100 SXM 80GB x1
A100 · 1× · 80GB VRAM · training
Training / fine-tuning
$1.64
/hour
Spot: $1.14/hr
A100 SXM 80GB x8
A100 · 8× · 80GB VRAM · training
Large-scale multi-GPU training
$13.12
/hour
Spot: $9.12/hr
H100 SXM 80GB x1
H100 · 1× · 80GB VRAM · training
Frontier training (availability varies)
$2.69
/hour
Spot: $1.99/hr
H100 SXM 80GB x8
H100 · 8× · 80GB VRAM · training
Full H100 pod
$21.52
/hour
Spot: $15.92/hr
Vast.ai logo
Vast.ai GPU Cloud
GPU Marketplace · Marketplace pricing across many GPU types.
US · EU · Asia

RTX 3090 24GB x1
RTX 3090 · 1× · 24GB VRAM · inference
Marketplace rate varies by host
varies
/hour
Spot: Market rate/hr
RTX 4090 24GB x1
RTX 4090 · 1× · 24GB VRAM · inference
Often strong value; prices vary
varies
/hour
Spot: Market rate/hr
L40S 48GB x1
L40S · 1× · 48GB VRAM · inference
Competitive rates on marketplace
varies
/hour
Spot: Market rate/hr
A100 40GB x1
A100 · 1× · 40GB VRAM · training
Training marketplace nodes
varies
/hour
Spot: Market rate/hr
A100 80GB x1
A100 · 1× · 80GB VRAM · training
Higher VRAM nodes (availability varies)
varies
/hour
Spot: Market rate/hr
H100 80GB x1
H100 · 1× · 80GB VRAM · training
Rare but possible on marketplace
varies
/hour
Spot: Market rate/hr
CoreWeave logo
CoreWeave GPU Cloud
GPU Cloud · Enterprise GPU cloud for large-scale AI.
from $0.26/hour
US · EU

RTX A4000 16GB x1
RTX A4000 · 1× · 16GB VRAM · inference
Budget workstation GPU for lighter jobs
$0.26
/hour
L40S 48GB x1
L40S · 1× · 48GB VRAM · inference
Strong inference / mid-tier training
$2.25
/hour
A100 80GB x1
A100 · 1× · 80GB VRAM · training
High VRAM training tier
$2.70
/hour
H100 80GB x1
H100 · 1× · 80GB VRAM · training
Frontier training
$6.16
/hour
H100 80GB x8
H100 · 8× · 80GB VRAM · training
Full H100 pod for enterprise training
$49.24
/hour
Paperspace logo
Paperspace GPU Cloud
Gradient · Simple GPU machines for developers.
from $0.76/hour
US · EU

T4 16GB x1
T4 · 1× · 16GB VRAM · inference
Budget inference / prototyping
$0.76
/hour
RTX A4000 16GB x1
RTX A4000 · 1× · 16GB VRAM · inference
General-purpose GPU option
$0.76
/hour
RTX A5000 24GB x1
RTX A5000 · 1× · 24GB VRAM · inference
More VRAM for bigger models
$1.38
/hour
A100 80GB x1
A100 · 1× · 80GB VRAM · training
Training tier (availability varies)
$1.15
/hour
OVHcloud logo
OVHcloud GPU Cloud
Bare Metal GPU · European GPU hosting options.
EU · US · Asia · Australia · Canada

T4 16GB x1
T4 · 1× · 16GB VRAM · inference
Entry GPU tier (varies by region/stock)
varies
/hour
RTX 4090 24GB x1
RTX 4090 · 1× · 24GB VRAM · inference
Offering varies by region/stock
varies
/hour
L40S 48GB x1
L40S · 1× · 48GB VRAM · inference
Offering varies by region/stock
varies
/hour
A100 80GB x1
A100 · 1× · 80GB VRAM · training
Offering varies by region/stock
varies
/hour
H100 80GB x1
H100 · 1× · 80GB VRAM · training
Offering varies by region/stock
varies
/hour
Crusoe Energy logo
Crusoe Energy GPU Cloud
Cloud GPU · Sustainable GPU cloud with on-demand, spot, and reserved options.
from $1.04/hour
US

L40S 48GB x1
L40S · 1× · 48GB VRAM · inference
Inference / mid-tier training
$1.04
/hour
A100 80GB x1
A100 · 1× · 80GB VRAM · training
Training — high VRAM
$2.42
/hour
A100 80GB x8
A100 · 8× · 80GB VRAM · training
Multi-GPU training cluster
$19.36
/hour
H100 80GB x1
H100 · 1× · 80GB VRAM · training
Frontier training
$3.90
/hour
Spot: $1.95/hr
H100 80GB x8
H100 · 8× · 80GB VRAM · training
Full H100 pod
$31.20
/hour
Spot: $15.60/hr
CloudMart AI
GPU
Find the right GPU cloud plan.
Describe your workload, VRAM needs, and budget to get a recommendation. You can also ask general questions like "H100 vs A100" or "how much VRAM does a 70B model need."
Compare:
Report a problem