GPU Cloud Pricing Comparison
Weekly-verified GPU rental prices across RunPod, Vast.ai, Lambda Labs, Hyperstack, CoreWeave, and Modal. Includes $ per GB-VRAM-hour — a normalized efficiency column nobody else publishes.
Affiliate links labeled — we may earn a commission at no cost to you.
| Provider | GPU | VRAM | $/hr on-demand ↑ | $/hr spot | $ per GB-VRAM-hr | Monthly (24/7 OD) | vCPU / RAM | Regions |
|---|---|---|---|---|---|---|---|---|
| Vast.ai ↗affiliate | RTX 3090 | 24 GB | $0.35 | $0.22 | $0.0146 | $255 | 8C / 32GB | global |
| RunPod ↗affiliate | RTX 3090 | 24 GB | $0.44 | $0.29 | $0.0183 | $321 | 8C / 32GB | us-east |
| Hyperstack ↗affiliate | RTX 4090 | 24 GB | $0.57 | — | $0.0237 | $416 | 16C / 64GB | eu-west |
| Vast.ai ↗affiliate | RTX 4090 | 24 GB | $0.59 | $0.39 | $0.0246 | $431 | 16C / 64GB | us, eu, asia |
| RunPod ↗affiliate | RTX 4090 | 24 GB | $0.74 | $0.49 | $0.0308 | $540 | 16C / 64GB | us-east, eu-west |
| Lambda Labs ↗affiliate | RTX 4090 | 24 GB | $0.75 | — | $0.0313 | $548 | 4C / 32GB | us-west |
| RunPod ↗affiliate | L40S 48GB | 48 GB | $1.25 | $0.89 | $0.0260 | $913 | 32C / 128GB | us-east, eu-west |
| RunPod ↗affiliate | A100 40GB SXM | 40 GB | $1.89 | $1.24 | $0.0473 | $1380 | 32C / 128GB | us-east |
| Hyperstack ↗affiliate | A100 80GB SXM | 80 GB | $2.06 | — | $0.0258 | $1504 | 32C / 160GB | eu-west |
| CoreWeave ↗affiliate | A100 80GB SXM | 80 GB | $2.21 | — | $0.0276 | $1613 | 30C / 200GB | us-east, us-west, eu |
| Vast.ai ↗affiliate | A100 80GB | 80 GB | $2.25 | $1.65 | $0.0281 | $1643 | 32C / 128GB | us, eu |
| RunPod ↗affiliate | A100 80GB SXM | 80 GB | $2.49 | $1.79 | $0.0311 | $1818 | 64C / 256GB | us-east, eu-west |
| Lambda Labs ↗affiliate | A100 80GB SXM | 80 GB | $2.49 | — | $0.0311 | $1818 | 30C / 200GB | us-west, us-east, eu |
| Hyperstack ↗affiliate | H100 80GB SXM | 80 GB | $2.85 | — | $0.0356 | $2081 | 26C / 200GB | eu-west |
| Vast.ai ↗affiliate | H100 80GB | 80 GB | $2.99 | $2.29 | $0.0374 | $2183 | 64C / 256GB | us, eu |
| Modal ↗affiliate | A100 80GB | 80 GB | $3.28 | — | $0.0410 | $2394 | 32C / 128GB | us |
| RunPod ↗affiliate | H100 80GB SXM | 80 GB | $3.29 | $2.49 | $0.0411 | $2402 | 64C / 256GB | us-east, eu-west |
| Lambda Labs ↗affiliate | H100 80GB SXM | 80 GB | $3.29 | — | $0.0411 | $2402 | 26C / 200GB | us-west, us-east |
| Modal ↗affiliate | H100 80GB | 80 GB | $4.02 | — | $0.0502 | $2935 | 32C / 128GB | us |
| CoreWeave ↗affiliate | H100 80GB SXM | 80 GB | $4.25 | — | $0.0531 | $3103 | 26C / 200GB | us-east, us-west, eu |
| CoreWeave ↗affiliate | H200 141GB | 141 GB | $5.25 | — | $0.0372 | $3833 | 52C / 400GB | us-east |
21 GPU listings shown. Click column headers to sort. Prices verified 2026-08-19.
How to Use
Filter by VRAM
Set the minimum VRAM to match your model's requirements. Use the LLM RAM Calculator to find how much VRAM your model needs.
Filter by provider
Select a specific GPU cloud provider or view all. Prices vary 2–3× for identical hardware across providers.
Sort by $ per GB-VRAM-hr
This derived column — unique to WiseCrewAI — shows the true cost efficiency of each listing. Lower is better.
Click to rent
Every row links directly to the provider. Links are labeled as affiliate links — we may earn a commission at no extra cost to you.
How to read this table
The $ per GB-VRAM-hr column normalizes cost by memory: an A100 40GB at $1.89/hr costs more per GB than an A100 80GB at $2.49/hr. Always use this column to compare cards with different VRAM sizes. The monthly (on-demand) column assumes 24/7 uptime (730 hours). Use the Self-Host vs API Calculator to see if self-hosting beats your current API spend.
FAQ
Have more questions? Contact us