Live cloud GPU rental prices across RunPod, the Vast.ai marketplace, DigitalOcean, Vultr, and GPU Mart, aggregated and normalized so you can spot the cheapest H100, A100, RTX 4090, or any other card in one view. Refreshed hourly.
One row per GPU with the cheapest current offer. Expand a GPU to compare every provider and tier, including spot. Click a GPU name to see which models fit.
| GPU | VRAM | Provider | Tier | Price / GPU-hr | Price / GPU-month | 30-day Trend | Rent |
|---|---|---|---|---|---|---|---|
NVIDIA GeForce RTX 3060 rtx-3060 | 12 GB | Vast.ai | Spot Cheapest Offer | $0.03 / GPU-hour | $19 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4060 rtx-4060 | 8 GB | Vast.ai | Spot Cheapest Offer | $0.03 / GPU-hour | $19 / GPU-month | Rent now | |
NVIDIA Tesla V100 16GB v100 | 16 GB | Vast.ai | Spot Cheapest Offer | $0.04 / GPU-hour | $29 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3060 Ti rtx-3060-ti | 8 GB | Vast.ai | Spot Cheapest Offer | $0.04 / GPU-hour | $29 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3080 rtx-3080 | 10 GB | Vast.ai | Spot Cheapest Offer | $0.04 / GPU-hour | $30 / GPU-month | Rent now | |
NVIDIA GeForce RTX 2080 Ti rtx-2080-ti | 11 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $34 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3070 Ti rtx-3070-ti | 8 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $37 / GPU-month | Rent now | |
NVIDIA RTX A4000 a4000 | 16 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $39 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4060 Ti rtx-4060-ti | 8 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $39 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5060 Ti rtx-5060-ti | 16 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $39 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4070 rtx-4070 | 12 GB | Vast.ai | Spot Cheapest Offer | $0.05 / GPU-hour | $39 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3070 rtx-3070 | 8 GB | Vast.ai | On-Demand Cheapest Offer | $0.06 / GPU-hour | $41 / GPU-month | Rent now | |
NVIDIA GeForce GTX 1660 Ti gtx-1660-ti | 6 GB | Vast.ai | Spot Cheapest Offer | $0.06 / GPU-hour | $44 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5060 rtx-5060 | 8 GB | Vast.ai | Spot Cheapest Offer | $0.07 / GPU-hour | $49 / GPU-month | Rent now | |
NVIDIA A16 a16 | 16 GB | Vast.ai | Spot Cheapest Offer | $0.07 / GPU-hour | $49 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3090 rtx-3090 | 24 GB | Vast.ai | Spot Cheapest Offer | $0.07 / GPU-hour | $52 / GPU-month | Rent now | |
NVIDIA GeForce RTX 3080 Ti rtx-3080-ti | 12 GB | Vast.ai | Spot Cheapest Offer | $0.08 / GPU-hour | $55 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4070 Ti rtx-4070-ti | 12 GB | Vast.ai | Spot Cheapest Offer | $0.08 / GPU-hour | $58 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4090 rtx-4090 | 24 GB | Vast.ai | Spot Cheapest Offer | $0.09 / GPU-hour | $66 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5070 Ti rtx-5070-ti | 16 GB | Vast.ai | Spot Cheapest Offer | $0.09 / GPU-hour | $68 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4080 rtx-4080 | 16 GB | Vast.ai | Spot Cheapest Offer | $0.11 / GPU-hour | $79 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5070 rtx-5070 | 12 GB | Vast.ai | On-Demand Cheapest Offer | $0.11 / GPU-hour | $81 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5080 rtx-5080 | 16 GB | Vast.ai | Spot Cheapest Offer | $0.13 / GPU-hour | $97 / GPU-month | Rent now | |
NVIDIA L4 l4 | 24 GB | Vast.ai | Spot Cheapest Offer | $0.13 / GPU-hour | $97 / GPU-month | Rent now | |
NVIDIA RTX A6000 a6000 | 48 GB | Vast.ai | Spot Cheapest Offer | $0.13 / GPU-hour | $97 / GPU-month | Rent now | |
NVIDIA RTX PRO 4000 Blackwell rtx-pro-4000 | 24 GB | Vast.ai | Spot Cheapest Offer | $0.13 / GPU-hour | $97 / GPU-month | Rent now | |
NVIDIA RTX PRO 5000 Blackwell rtx-pro-5000 | 48 GB | Vast.ai | Spot Cheapest Offer | $0.14 / GPU-hour | $102 / GPU-month | Rent now | |
NVIDIA RTX A5000 a5000 | 24 GB | RunPod | Community | $0.16 / GPU-hour | $117 / GPU-month | Rent now | |
NVIDIA GeForce RTX 5090 rtx-5090 | 32 GB | Vast.ai | Spot Cheapest Offer | $0.19 / GPU-hour | $137 / GPU-month | Rent now | |
NVIDIA L40S l40s | 48 GB | Vast.ai | Spot Cheapest Offer | $0.20 / GPU-hour | $146 / GPU-month | Rent now | |
NVIDIA RTX 4000 Ada rtx-4000-ada | 20 GB | RunPod | Community | $0.20 / GPU-hour | $146 / GPU-month | Rent now | |
Apple A10 Fusion GPU apple-a10 | 3 GB | Vast.ai | On-Demand Cheapest Offer | $0.20 / GPU-hour | $147 / GPU-month | Rent now | |
NVIDIA A100 80GB SXM a100-80gb-sxm | 80 GB | Vast.ai | Spot Cheapest Offer | $0.24 / GPU-hour | $175 / GPU-month | Rent now | |
NVIDIA A100 80GB PCIe a100-80gb-pcie | 80 GB | Vast.ai | Spot Cheapest Offer | $0.24 / GPU-hour | $177 / GPU-month | Rent now | |
NVIDIA RTX PRO 2000 Blackwell rtx-pro-2000 | 16 GB | GPU Mart | Monthly | $0.25 / GPU-hour | $179 / GPU-month Annual term: $159/mo | No history yet | Rent now |
NVIDIA L40 l40 | 48 GB | Vast.ai | Spot Cheapest Offer | $0.27 / GPU-hour | $195 / GPU-month | Rent now | |
NVIDIA GeForce RTX 4080 SUPER rtx-4080-super | 16 GB | RunPod | Community | $0.28 / GPU-hour | $204 / GPU-month | Rent now | |
NVIDIA A40 a40 | 48 GB | RunPod | Community | $0.35 / GPU-hour | $255 / GPU-month | Rent now | |
AMD Instinct MI300X mi300x | 192 GB | RunPod | Community | $0.50 / GPU-hour | $365 / GPU-month | Rent now | |
NVIDIA H200 NVL h200-nvl | 141 GB | RunPod | Community | $0.50 / GPU-hour | $365 / GPU-month | Rent now | |
NVIDIA H100 SXM h100-sxm | 80 GB | Vast.ai | Spot Cheapest Offer | $0.67 / GPU-hour | $487 / GPU-month | Rent now | |
NVIDIA H100 PCIe h100-pcie | 80 GB | Vast.ai | Spot Cheapest Offer | $0.73 / GPU-hour | $535 / GPU-month | Rent now | |
NVIDIA RTX 6000 Ada rtx-6000-ada | 48 GB | RunPod | Community | $0.74 / GPU-hour | $540 / GPU-month | Rent now | |
NVIDIA A100 40GB PCIe a100-40gb-pcie | 40 GB | GPU Mart | Monthly | $0.86 / GPU-hour | $625 / GPU-month Annual term: $525/mo 4-GPU server, $2,499/mo total | No history yet | Rent now |
NVIDIA RTX PRO 6000 Blackwell rtx-pro-6000 | 96 GB | GPU Mart | Monthly | $1.09 / GPU-hour | $799 / GPU-month Annual term: $699/mo | No history yet | Rent now |
NVIDIA H200 SXM h200-sxm | 141 GB | Vast.ai | Spot Cheapest Offer | $1.32 / GPU-hour | $961 / GPU-month | Rent now | |
NVIDIA H100 NVL h100-nvl | 94 GB | RunPod | Community | $2.59 / GPU-hour | $1,891 / GPU-month | Rent now | |
AMD Instinct MI325X mi325x | 256 GB | DigitalOcean | On-Demand | $3.8 / GPU-hour | $2,774 / GPU-month | Rent now | |
AMD Instinct MI350X mi350x | 288 GB | DigitalOcean | Spot | $4 / GPU-hour | $2,920 / GPU-month | Rent now | |
AMD Instinct MI355X mi355x | 288 GB | DigitalOcean | Spot | $4.5 / GPU-hour | $3,285 / GPU-month | Rent now | |
NVIDIA B200 b200 | 192 GB | Vast.ai | Spot Cheapest Offer | $5 / GPU-hour | $3,650 / GPU-month | Rent now | |
NVIDIA B300 b300 | 288 GB | RunPod | Community | $6.94 / GPU-hour | $5,066 / GPU-month | Rent now |
Methodology: Pulled live each hour from the RunPod, Vast.ai, DigitalOcean, and Vultr pricing APIs, normalized to per-GPU USD/hr. Vast.ai figures cover verified hosts only. Vultr fractional GPU slices and out-of-stock plans are left out. GPU Mart prices come from their published rate card and are updated by hand. No affiliate weighting. GPU Mart bills monthly. Its hourly price is the monthly rate divided by 730 hours. Monthly figures for RunPod and Vast.ai are the hourly rate times 730, so they assume the instance runs around the clock.
Limited-Time Offer
Sign up through our referral link and Vultr adds $300 in credit to your new account. That is enough to run a whole cloud GPU for days before you commit to anything.
Referral link. We earn a commission if you become a paying customer, at no cost to you.
The rental cards people search for most, with the cheapest live per-GPU price we track right now. Jump to the full table for every tier and provider.
From $1.73/hr on Vast.ai
The default datacenter GPU for training and heavy inference. 80GB of HBM3 fits 70B parameter models without sharding.
See all listingsFrom $0.43/hr on Vast.ai
Still the price-performance pick for fine-tuning and long training runs. Wide marketplace supply keeps hourly rates low.
See all listingsFrom $0.23/hr on Vast.ai
The consumer workhorse for inference. 24GB of VRAM runs most quantized 32B models at interactive speed.
See all listingsFrom $0.37/hr on Vast.ai
The fastest consumer card we track. 32GB of VRAM and GDDR7 bandwidth make it a strong single-GPU inference rig.
See all listingsA cloud GPU rental is an hourly-billed compute instance with a single dedicated GPU (or a small group), priced in US dollars per GPU-hour. This index pulls the cheapest current rate for every common AI-workload GPU directly from the public APIs of RunPod, the Vast.ai marketplace, DigitalOcean, and Vultr, adds GPU Mart’s monthly dedicated servers as an hourly equivalent, normalizes the SKU names, and refreshes every hour.
Coverage spans NVIDIA H100, H200, A100, B200, L40S, RTX 6000 Ada, RTX 5090, RTX 4090, RTX 3090, and roughly 30 other AI-relevant cards across multiple regions and tiers. For each card we surface the cheapest on-demand rate and, where it applies, a 25th-percentile floor so a single mispriced Vast.ai listing cannot anchor the headline.
Tiers are explicit: Secure (vetted Tier-3 or Tier-4 datacenter), Community (peer-operated host on the same provider), Spot (interruptible, cheapest), On-Demand (standard non-interruptible), and Monthly (a dedicated server billed by the month, divided by 730 hours). All prices are quoted per GPU-hour so multi-GPU listings are directly comparable to single-GPU ones. A separate daily snapshot powers the 30-day trend sparkline next to each row.
Pair With
Plug these hourly rental rates into the three-way decision tool to see whether renting beats buying or calling an API at your usage volume.
Open the Decision Tool
Per-token input and output prices for 350+ hosted models from OpenRouter.
Find the break-even between buying a GPU and paying for a cloud API.
Browse GPUs, edge devices, TPUs, and AI PCs with specs and real prices.
Budget-matched workstation builds for the GPUs in this index.
Straight answers to the questions we hear most often.
Can't find what you're looking for? Send us a message
The live table refreshes hourly. We call each provider’s public API directly, normalize SKU names to canonical GPU specs, and cache the result for one hour so a high-traffic moment never overwhelms the upstream services. A separate daily snapshot records prices for the 30-day trend chart.
Either the provider does not stock that GPU, or its public API uses an SKU name we have not mapped yet. Unmapped SKUs are logged on every sync so we can extend the mapping. If a card you expect is missing, get in touch.
Secure refers to vetted Tier-3 or Tier-4 datacenter hosts. Community refers to lower-cost hosts on the same provider, often peer-operated. Spot is interruptible compute, the cheapest tier but can be reclaimed by the provider with little notice. On-demand is the standard non-interruptible price.
Per GPU. Where a provider quotes a multi-GPU instance, we divide by the GPU count so prices are directly comparable across single and multi-GPU listings.
Prices vary by region, host reliability, and availability. We surface the cheapest currently rentable per-GPU rate from each provider’s public marketplace or pricing API. The provider page might show a different default region or a non-rentable listing.
Vast.ai exposes many independent host listings per GPU model. We capture two snapshots per group, the absolute minimum and the 25th percentile, so a single outlier listing cannot misrepresent the price. The "min" row is what surfaces as the headline.
As of mid-2026, a single H100 SXM runs roughly $2 to $3 per GPU-hour on-demand on RunPod and Vast.ai, with spot listings dipping lower. The live table above always shows the current cheapest rate we track, refreshed hourly, so check it rather than trusting any static number.
It depends on the card. Vast.ai is a marketplace of independent hosts and usually undercuts on consumer GPUs like the RTX 4090 and 5090. RunPod is often more competitive on datacenter cards and offers vetted secure-tier hosts. DigitalOcean and Vultr publish fixed list prices, so they rarely win on price but you get a predictable bill and a mainstream cloud around the GPU. GPU Mart bills by the month, so it only wins when a server runs around the clock. Use the monthly column to compare on that basis.
Renting wins for bursty or short-term workloads; buying wins once utilization is high and sustained. As a rule of thumb, if you would rent the card for more than a few hours a day for a year, buying usually pays off. Our self-host vs rent vs API decision tool runs the break-even math with these live prices.
They are on the roadmap. Lambda Labs requires an account-tied API key that is gated for new signups, and CoreWeave, Modal, and Colab do not expose clean public pricing APIs. RunPod, Vast.ai, DigitalOcean, and Vultr together cover the bulk of practical price comparison for self-serve renters, and we will add more sources as access opens up.
