What is the cheapest GPU cloud provider?

Pricing varies daily by GPU model and billing type — use our Cost Calculator to compare current rates.

How much does it cost to rent an H100 GPU?

H100 pricing varies by provider and billing type — check our H100 pricing page for current rates.

What GPU do I need to run an LLM locally?

It depends on model size and quantization — use our Workload Recommender to find compatible GPUs.

How does nodepedia collect pricing data?

An AI agent visits provider pricing pages daily and extracts current rates automatically.

Can I compare GPU cloud providers side by side?

Yes — nodepedia offers head-to-head comparisons for every provider pair.

What is the difference between spot and on-demand GPU pricing?

Spot instances use spare capacity at a discount but can be interrupted; on-demand guarantees availability at a premium.

Is nodepedia free to use?

Yes — all pricing data, tools, and guides are completely free with no signup required.

All Providers Price Comparison GPUs Models Comparisons Tools Glossary Guides Search

Compare

Independent GPU cloud pricing, updated daily. 28 providers, 86+ GPUs compared. Free, no signup.

Compute Calculator Browse Providers Browse GPUs

GPU Providers

GPU Models

857

LLMs

Daily

Price Updates

Popular GPUs

All GPUs →

GPU Model	VRAM	Providers	From
A100 80GB	80GB	18	$0.33 /hr
H100 SXM 80GB	80GB	17	$0.60 /hr
H200 SXM 141GB	141GB	16	$1.49 /hr
L40S 48GB	48GB	15	$0.40 /hr
Blackwell B200	192GB	13	$1.77 /hr
A100 40GB	40GB	12	$0.45 /hr
L4	24GB	11	$0.13 /hr
RTX A6000	48GB	11	$0.21 /hr

Popular Models

All Models →

Model	Developer	Released	Params	Min VRAM	Compatible GPUs
granite-4.1-8b	ibm-granite	Apr 2026	8.79B	5.8GB	81
karma-electric-llama31-8b	anicka	Feb 2026	8.03B	5.3GB	81
LFM2.5-8B-A1B	LiquidAI	May 2026	8.47B	5.59GB	81
Nemotron-Labs-Diffusion-8B	nvidia	Mar 2026	8.49B	5.6GB	81
NVIDIA-Nemotron-Nano-9B-v2-Japanese	nvidia	Feb 2026	8.89B	5.86GB	81
Qwen3.5-9B-abliterated	lukey03	Mar 2026	8.95B	5.91GB	81
steerling-8b	guidelabs	Feb 2026	8.39B	5.54GB	81
gemma-4-E4B-it-OBLITERATED	OBLITERATUS	Apr 2026	8.00B	5.28GB	81
gemma4-e4b-claims-comparison	k-chirkunov	Jul 2026	7.94B	5.24GB	81
Olmo-Hybrid-Instruct-DPO-7B	allenai	Feb 2026	7.43B	4.91GB	81

Live Feed

1 / 12

Provider

HOSTKEY

GPUs

From

$0.32/hr

Verified 2026-07-31

Price Alert

A100 80GB

Cheapest

$0.33/hr

Via

Vast.ai

VRAM

80GB

Providers

Updated 2026-08-03

LLM

DeepSeek-V4-Flash-0731

Developer

deepseek-ai

Params

304.180418494B

Compatible GPUs

Released 2026-07-31

Provider

Vast.ai

GPUs

From

$0.01/hr

Verified 2026-08-03

Price Alert

H100 SXM 80GB

Cheapest

$0.60/hr

Via

Vast.ai

VRAM

80GB

Providers

Updated 2026-08-03

LLM

base-grpo-round1

Developer

sullivan1502

Params

0.033588224B

Compatible GPUs

Released 2026-07-27

Provider

Clore.ai

GPUs

From

$0.01/hr

Verified 2026-08-03

Price Alert

H200 SXM 141GB

Cheapest

$1.49/hr

Via

Verda

VRAM

141GB

Providers

Updated 2026-08-03

LLM

Kimi-K3-DSpark

Developer

RadixArk

Params

2.249289601B

Compatible GPUs

Released 2026-07-27

Provider

RunPod

GPUs

From

$0.12/hr

Verified 2026-08-03

Price Alert

L40S 48GB

Cheapest

$0.40/hr

Via

Vast.ai

VRAM

48GB

Providers

Updated 2026-08-03

Provider

CUDO Compute

GPUs

From

$0.35/hr

Verified 2026-08-02

GPU Cloud Pricing

All Prices →

GPU Model	VRAM	Atlas Cloud	Bitdeer AI	Cantech	Clore.ai	Hivenet Compute	CUDO Compute	DigitalOcean	E2E Networks	GMI Cloud	GPU.ai	HOSTKEY	Hostrunway	Hyperstack	Jarvis Labs	Nebius AI	Northflank	Packet AI	DigitalOcean	Qubrid AI	RedSwitches	RunPod	Spheron Network	SynpixCloud	TensorDock	Theta Edge Cloud	Thunder Compute	Vast.ai	Verda
A100 80GB	80GB	—	—	—	—	—	$1.35 –	$3.44^* ↓	$2.1^* –	—	$1.46 ↓	$5.28^* ↑	—	$1.35^* –	$0.89 –	—	$3.78^* ↑	$1.43 –	$3^* –	$1.62^* –	—	$1.39 –	$0.82 –	$3.12^* ↑	$0.85^* –	—	$0.84 ↑	$0.33 ↓	$0.63 ↑
H100 SXM 80GB	80GB	—	$1.97 ↓	—	—	—	—	$7.69^* ↑	$3.77^* –	$2^* –	$2.83 –	$14.07^* ↑	—	$2.5^* ↓	$1.19 –	$2.15 –	$2.74^* ↓	$3.75^* ↑	$7.5^* –	$4.78^* –	—	$2.69 –	$1.77 ↑	$2.79^* –	—	—	$2.25^* ↑	$0.6 ↑	$1.14 –
H200 SXM 141GB	141GB	$3.5^* –	$2.26 ↓	—	—	—	—	$12.94^* ↑	$4.54^* –	$2.6^* –	$4.19 –	$9.71^* ↑	—	$3.99^* –	$14.1^* –	$2.45 –	—	$5.9^* ↑	—	$5.56^* –	—	$3.59^* –	$1.77 ↓	—	—	$2.49^* –	—	$2.63 ↑	$1.49 –
L40S 48GB	48GB	—	—	—	$0.5^* ↓	—	$0.87 –	$3.17^* ↑	$1.2^* –	—	$1.1 –	$4.1^* ↑	—	—	—	$0.74 –	—	$0.92 –	$2.25^* –	$2.48^* –	$1.06^* ↓	$0.79 –	$0.61 ↓	—	—	—	—	$0.4 ↓	$0.48 –
Blackwell B200	192GB	—	$4.8 –	—	—	—	—	—	$6.99^* –	$4^* –	$6.19 –	$12.96^* ↑	—	$6^* –	$10^*	$3.95 –	—	$3.75 –	—	$5.63^* –	—	$5.98^* –	$1.77 ↓	—	—	—	—	$5 ↓	$2.14 –
A100 40GB	40GB	—	—	$1.13^* –	—	—	—	—	$1.98^* –	—	$1.45 –	$3.23^* ↑	—	$1.35^* ↑	$0.79 –	—	$1.75^* ↑	—	—	$1.38^* –	—	$2.74 –	—	$0.73^* ↓	$0.85^* –	—	—	—	$0.45 –
L4	24GB	—	—	—	—	—	—	—	$0.57^* –	—	$0.42 –	$1.24^* ↑	$0.88^* –	—	$0.29 –	—	$0.65^* ↑	—	—	$1.11^* –	$0.59^* –	$0.44^* –	—	—	$0.23^* –	—	—	$0.13 –	—
RTX A6000	48GB	—	—	—	$0.35^* –	—	$0.45^* –	—	—	—	$0.79 –	$6.21^* ↑	—	$0.5^* ↑	—	—	—	—	—	—	—	$0.33 –	—	$0.65^* ↓	$0.39^* –	—	$0.35^* –	$0.27 ↓	$0.21 –
RTX 5090	32GB	—	—	—	$0.15^* –	$0.81^* –	—	—	—	—	—	$6.91^* ↑	—	—	—	—	—	$0.66 ↓	—	—	—	$0.69^* –	$0.53 ↓	$1.6^* ↑	$0.54^* –	$0.29^* ↓	—	$0.14 ↑	—
H100 PCIe 80GB	80GB	—	—	—	—	—	—	—	—	—	$2.49 –	$7.55^* ↑	—	$2 ↓	$3.99^* ↑	—	—	—	—	—	—	$1.99 –	$1.77 ↑	$2.79^* ↓	$1.99^* –	—	$1.09 ↓	$0.35 ↓	$1.14 –
RTX 6000 Ada Generation	48GB	—	—	—	—	—	—	$4.74^* ↑	$1.9^* –	—	$0.95 –	$8.63^* ↑	—	—	$0.99 –	—	$3.57^* ↓	—	—	—	—	$0.74 –	—	—	$0.7^* –	—	—	$0.13 –	$0.36 –
RTX 6000 Ada Generation	96GB	—	—	—	—	—	—	—	$1.9^* –	—	—	$7.02^* ↑	—	$1.85^* ↑	$1.89^* –	$0.95 –	$3.67^* –	—	—	—	—	$1.69^* –	$0.64 –	—	—	—	—	$0.4 ↑	$0.66 –
RTX 4090	24GB	—	—	—	$0.01^* –	$0.43^* –	—	—	—	—	$0.44 –	$5.73^* ↑	—	—	—	—	—	$0.39 ↓	—	—	—	$0.34 –	—	$0.48^* –	$0.38^* –	—	—	$0.11 ↑	—
RTX 3090	24GB	—	—	—	$0.03^* –	—	—	—	—	—	—	$3.39^* ↑	—	—	—	—	—	$1.99^* –	—	—	—	$0.22 –	—	$0.32^* ↓	$0.2^* –	$0.12 ↓	—	$0.07 ↑	—
Tesla V100	32GB	—	—	—	—	—	$0.39^* –	—	—	—	$0.29 –	$1.94^* ↑	—	—	—	—	—	—	—	—	—	$0.19 –	—	$0.32^* ↓	$0.19^* –	—	—	$0.07 ↑	$0.06 –
RTX A5000	24GB	—	—	—	$0.19^* –	—	$0.35^* –	—	—	—	—	$3.72^* ↑	—	—	—	—	—	—	—	—	—	$0.16 –	—	$0.47^* –	$0.21^* –	—	—	$0.03 –	—
T4	16GB	—	—	—	$0.1^* –	—	—	$0.6^* –	—	—	—	$0.76^* ↓	$0.71^* –	—	—	—	—	—	—	$0.73^* –	$0.44^* –	—	—	—	—	—	—	$0.05 –	—
A100 PCIe 80GB	80GB	—	—	—	—	—	$1.5 –	—	—	—	—	$3.02^* ↓	—	—	—	—	—	—	—	—	—	$1.19 –	—	$1.58^* –	$1.5^* –	—	—	$0.13 –	—
A40	48GB	—	—	$0.13^* –	—	—	$0.39^* –	—	$1.44^* –	—	—	$3.24^* ↑	—	—	—	—	—	—	—	—	—	$0.3 –	—	$0.63^* –	—	—	—	$0.31 ↑	—
RTX 3080	10GB	—	—	—	$0.03^* –	—	—	—	—	—	—	$2.25^* ↑	—	—	—	—	—	$1.49^* –	—	—	—	$0.17 –	—	$0.23^* –	—	—	—	$0.04 ↓	—

All prices are per hour. Spot prices shown where available — prices marked with ^* are on-demand. The lowest price in each row is highlighted.

Data verified daily 28 GPU cloud providers 86 GPU models Independent pricing

Why Compare GPU Cloud Pricing?

GPU cloud pricing changes daily. Spot prices fluctuate hourly. The same GPU can vary by 2–3x between providers depending on availability and billing type. As of August 2026, nodepedia tracks real pricing across 28 GPU cloud providers and 86 GPU models so you can find the cheapest option for your workload — try the Cost Calculator to estimate your spend.

What You Can Do

Compare spot and on-demand pricing across all major providers
Check which GPUs can run your LLM at different quantization levels
Estimate costs with the calculator before committing to a provider
Read head-to-head provider comparisons with real pricing data

New to cloud GPUs? Explore our GPU Cloud Glossary for plain-English explanations of key concepts like GPU nodes, VRAM, spot pricing, and more.

How Data Is Collected

An AI agent extracts pricing from provider websites daily. No data is self-reported by providers. Every price is pulled directly from the source, validated against historical patterns, and flagged if it looks anomalous. You get the same prices you'd see if you visited each provider yourself — just all in one place.

Who Uses nodepedia?

ML engineers comparing cloud options for training runs. Startups evaluating which GPU provider fits their budget. Researchers who need a specific GPU and want to find the lowest price. Anyone renting cloud GPUs who wants to stop overpaying by checking one site instead of a dozen.

Frequently Asked Questions

What is the cheapest GPU cloud provider?: Pricing changes daily and depends on the GPU model, billing type (spot vs on-demand), and availability. There is no single cheapest provider — it varies by workload. Use the Cost Calculator to compare current pricing across all 28 providers tracked by nodepedia.
How much does it cost to rent an H100 GPU?: H100 pricing varies significantly by provider and billing type. Spot instances are cheaper but can be interrupted, while on-demand instances guarantee availability at a premium. Check the H100 pricing page for current rates across all tracked providers.
What GPU do I need to run an LLM locally?: It depends on the model size, quantization level, and whether you need training or inference. A 7B parameter model at Q4 quantization fits on a 6 GB GPU, while a 70B model may need 40+ GB of VRAM. Use the Workload Recommender to match your model to compatible GPUs.
How does nodepedia collect pricing data?: An AI agent visits provider pricing pages daily and extracts current rates automatically. Prices are not self-reported by providers. Every data point is pulled directly from the source, validated against historical patterns, and flagged if anomalous.
Can I compare GPU cloud providers side by side?: Yes. nodepedia offers head-to-head comparisons for every provider pair, covering pricing, GPU availability, and billing options. You can also build custom comparisons with the Comparison Builder tool.
What is the difference between spot and on-demand GPU pricing?: Spot instances use spare GPU capacity at a discount but can be interrupted when demand rises. On-demand instances guarantee availability at a higher price. Spot pricing can be 50–80% cheaper, making it ideal for fault-tolerant workloads like training with checkpoints. The Cost Calculator shows both pricing types for easy comparison.
Is nodepedia free to use?: Yes. All pricing data, tools, and guides on nodepedia are completely free with no signup required. All data is independently collected — rankings and pricing are never influenced by providers.