Best GPU cloud providers (2026)

Rent GPUs by the hour for training, fine-tunes and inference. Supply moves the real price. We cite a public floor when the vendor publishes one, and we do not invent stock.

Updated Nov 2026 · 10 providers · 1 with a published monthly plan · 2 with a free tier

Best GPU cloud right now

Three starting points from the ranking below, each for a different kind of buyer.

Best free tier

Google Cloud

Free tier, then usage

83

The highest-ranked provider with a free plan or trial, after the picks above. Teams that want GCP data services or GPUs next to the app.

Visit Google Cloud

Full GPU cloud ranking

Sorted by score out of 100. A from-price is the cheapest plan the vendor publishes, not a quote for your workload.

  1. 01
    Amazon Web Services

    Lightsail from $3.50/mo

    83

    Teams that already live in AWS or need its service catalog

    Free tierVisit
  2. 02
    Google Cloud

    Free tier, then usage

    83

    Teams that want GCP data services or GPUs next to the app

    Free tierVisit
  3. 03
    CoreWeave

    Public GPU price list

    73

    Teams that need a lot of GPU, not one cheap hour

    No free tierVisit
  4. 04
    Crusoe

    Public GPU cloud rates

    73

    GPU capacity when site and energy story matter

    No free tierVisit
  5. 05
    Hyperstack

    Hourly GPU VMs

    73

    A single GPU VM without a hyperscaler account

    No free tierVisit
  6. 06
    Lambda

    Public on-demand GPU rates

    73

    On-demand training boxes with a known GPU

    No free tierVisit
  7. 07
    Paperspace

    Public hourly GPU rates

    73

    Notebooks and small training runs on a known platform

    No free tierVisit
  8. 08
    RunPod

    Community GPUs from about $0.20/hr

    73

    Fine-tunes and batch jobs that can checkpoint

    No free tierVisit
  9. 09
    TensorDock

    Marketplace hourly rates

    73

    A short GPU job when you are willing to pick the host

    No free tierVisit
  10. 10
    Vast.ai

    Marketplace, often from about $0.10/hr

    73

    The lowest GPU hour when you can tolerate a marketplace

    No free tierVisit

How GPU cloud pricing works

You pay for GPU hours, by model and by how the capacity is reserved. These are the lines that decide the invoice.

GPU model

The hourly rate depends mostly on the card: consumer cards, previous-generation datacenter cards and the latest accelerators sit in very different price bands.

On-demand, spot or reserved

On-demand is the list price. Spot or interruptible capacity is cheaper and can be taken back. Reserved clusters need a contract and usually a minimum term.

Storage and data transfer

Datasets, checkpoints and container images sit on persistent storage billed per GB per month. Moving data out costs extra on many clouds.

Idle time

A GPU bills while it is allocated, not while it computes. Notebooks left open overnight are a classic budget leak.

ProviderStarting priceBilling modelFree tier
Amazon Web ServicesLightsail from $3.50/moPay for what you use: requests, tokens, traffic or compute time. The bill moves with volume.Yes
Google CloudFree tier, then usagePay for what you use: requests, tokens, traffic or compute time. The bill moves with volume.Yes
CoreWeavePublic GPU price listBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
CrusoePublic GPU cloud ratesBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
HyperstackHourly GPU VMsBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
LambdaPublic on-demand GPU ratesBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
PaperspacePublic hourly GPU ratesBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
RunPodCommunity GPUs from about $0.20/hrBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
TensorDockMarketplace hourly ratesBilled by the hour for the resources you keep allocated. Idle capacity still counts.No
Vast.aiMarketplace, often from about $0.10/hrBilled by the hour for the resources you keep allocated. Idle capacity still counts.No

What fits

Marketplace GPU hour

You want the lowest published hourly floor and can tolerate variable stock or interrupts. RunPod, Vast.ai and TensorDock are typical starting points.

Steady inference node

You need on-demand capacity that stays up. Prefer hyperscaler GPU families or vendors that sell non-spot instances you can keep attached.

Training cluster

Multi-GPU interconnect and longer jobs matter. Shortlist CoreWeave, Crusoe, Lambda and hyperscaler training shapes; confirm the exact SKU and network before you schedule the run.

API instead of a GPU

Utilization is low or ops cost dominates. Buy tokens on an [LLM API](/llm-apis) until the GPU would stay busy enough to justify renting hardware.

How to choose a GPU cloud

GPU lists go stale quickly because stock and spot prices move. RunPod, Vast.ai, Lambda, TensorDock and Hyperstack publish hourly prices a developer can read without a quote. Amazon Web Services and Google Cloud publish instance families and add the usual cloud extras—storage, networking and idle disks. A marketplace floor is not a reserved contract. A reserved contract is not a spot interrupt.

Match the GPU to the job before you compare logos. Inference on a small model does not need an H100 node. Multi-GPU training cares about interconnect and whether the cluster stays up. CoreWeave, Crusoe and Paperspace sit in different parts of that spectrum: some lean toward large training estates, others toward notebooks and shorter jobs. Confirm VRAM, GPU count and the interrupt policy on the vendor page the day you buy.

Storage and egress often dominate short jobs. Pulling a large image or dataset once can cost more than the GPU hour. Check local disk size, network volume pricing and whether checkpoints land on object storage you already pay for elsewhere. If you only need a model behind an API key, compare LLM APIs first—running your own stack on a GPU makes sense when utilization stays high enough to beat token billing.

Community and spot capacity is cheaper because the machine can go away. That is fine when you checkpoint and can resume. It is a weak base for a latency SLA. On-demand or reserved capacity costs more and is the right comparison for a steady inference service. Do not mix those columns on a spreadsheet.

This ranking scores published pricing, a free credit when one exists, and self-serve access. It is not a tokens-per-second lab and it does not claim live stock. If you still need a plain CPU box for the control plane or a bastion, keep a VPS or cloud hosting account beside the GPU vendor instead of forcing every service onto an accelerator node.

Start with one paid hour on the image you will actually use. Driver, CUDA and the clock you were quoted should match the box you got. If the template fails to boot cleanly, that vendor is not “almost right”—it is the wrong shortlist for your stack.

  1. Step 1

    Name the GPU and the VRAM

    Compare that SKU. A provider that is cheap on an old card can be expensive on the card you need.

  2. Step 2

    Decide interruptible or not

    Spot and community clouds are cheaper because the machine can go away. Fine if you checkpoint. Bad if you cannot.

  3. Step 3

    Price storage and egress

    Datasets and image pulls can cost more than the GPU hour on a short job. Check both.

  4. Step 4

    Start on one hour

    Boot the image you will actually use. Driver, CUDA and the clock you were quoted should match the box you got.

Checklist before you buy

  • A named GPU with a public hourly price
  • VRAM and whether the card is shared
  • Interrupt policy: on-demand, spot or community
  • Per-second or per-minute billing, stated
  • Region and data-transfer price
  • A template or image that boots without a custom driver hunt

Common mistakes

Treating an hourly floor as a monthly price

A cheap hourly rate on a marketplace says nothing about availability or what a month of training costs. Estimate hours from your own runs.

Ignoring interconnect for multi-GPU jobs

Distributed training depends on the network between cards. Ask what interconnect the cluster uses before you compare price per GPU.

Relying on spot for deadlines

Interruptible capacity is fine for experiments with checkpoints. It is a risk for anything with a delivery date.

Forgetting where the data lives

Training on one cloud with data on another adds transfer cost and latency.

Compare GPU cloud head to head

How we rank GPU cloud

Each provider is scored out of 100 against the other 9providers on this page, using only what is published on the vendor's site. The sponsor slot is not an input.

Full editorial policy
35
Monthly floorWhere the cheapest published paid plan sits among peers in the same category. Cheaper scores higher. A vendor without a monthly floor gets the middle, so a missing price never counts as $0.
25
Public priceA rate card, plan grid or free tier you can read without talking to sales.
15
Self-serve signupYou can create an account and launch the resource yourself.
15
Documented scopeUp to three concrete, documented capabilities on the profile. Logos and ad spend are not counted.
10
Free tier or trialA bonus when the vendor publishes one. A free plan is a test, not a production price.

GPU Cloud glossary

On-demand
Capacity billed by the hour at list price, with no commitment and no guarantee it is available.
Spot / interruptible
Discounted capacity that the provider can reclaim with little notice.
Reserved cluster
A set of GPUs held for you for a contracted period, usually weeks to years.
VRAM
Memory on the GPU. It limits the model size and batch size you can fit on one card.
Interconnect
The link between GPUs and nodes. It matters for multi-GPU training more than for single-card inference.

GPU Cloud FAQ

Who has the cheapest GPU?

Marketplace clouds often show the lowest hourly floor, with variable stock and interrupt risk. Hyperscalers publish a higher, more stable on-demand price. Compare the same GPU, not the smallest number on the homepage.

Do you measure tokens per second?

No. DrDeploy does not publish lab benchmarks. We rank public commercial terms and say so.

Can I use a GPU cloud for a steady API?

Yes if you pick on-demand capacity and a region that stays in stock. Community floors are a weak base for a latency promise.

When is an LLM API better?

When you want a key and a rate card instead of drivers and utilization math. See LLM APIs. Rent GPUs when you run the weights yourself and can keep the cards busy.

Do prices include storage?

Often not. Local disk, network volumes and egress are separate lines. Price a real dataset pull, not only the GPU hour.

Is availability guaranteed?

Not on marketplace or spot inventory. Treat public floors as commercial facts for the SKU, then confirm stock when you are ready to buy.