Skip to content
GPU Finder

Explore the market

GPU Cloud Providers

Compare GPU cloud providers by NVIDIA model coverage, European regions, billing increments, price, and availability signals.

Every card shows the GPU models and configurations GPU Finder tracks, the current 1× on-demand price range, the cheapest H100 $/GPU-hour, a 7-day in-stock history for the 27 providers with a live stock source, and the recorded egress fee.

Cloud providers
35
With stock signals
27
Sort by
  • Vast.ai logo

    Vast.ai

    Marketplace

    Verified hosts with at least 90% reliability, 16 GB disk and 7 days remaining rental time.

    57 GPU models · 2,072 configurations · per-second billing

    $0.06 — $8.75/hr · 1× GPU

    Cheapest H100: $2.44/GPU-hr

    Free egress

    Live stock

    Top GPUs: RTX 5090 · RTX 4090 · RTX 3090 · RTX PRO 6000

    Explore provider
  • Runpod logo

    Runpod

    Specialist GPU cloud

    Specialist GPU cloud with Secure and Community tiers, per-second billing and live stock.

    37 GPU models · 227 configurations · per-second billing

    $0.13 — $8.99/hr · 1× GPU

    Cheapest H100: $2.59/GPU-hr

    Free egress

    Live stock

    Top GPUs: H100 · RTX PRO 6000 · A100-80GB · H200

    Explore provider
  • EmpirioLabs AI logo

    EmpirioLabs AI

    Specialist GPU cloud

    US inference provider with per-second GPU instances from RTX to B300; live unit counts from its public API.

    22 GPU models · 76 configurations · per-second billing

    $0.25 — $8.70/hr · 1× GPU

    Egress unknown

    Live stock

    Top GPUs: RTX PRO 6000 · H200 · H100 · B200

    Explore provider
  • Shadeform logo

    Shadeform

    Marketplace

    Multi-cloud GPU marketplace that books capacity across partner clouds; live stock via its API.

    22 GPU models · 203 configurations

    $0.57 — $5.99/hr · 1× GPU

    Cheapest H100: $1.99/GPU-hr

    Egress unknown

    Live stock

    Top GPUs: RTX A6000 · A100-80GB · H100 · A100

    Explore provider
  • AWS logo

    AWS

    Hyperscaler

    Hyperscaler with the broadest GPU catalog and region list; enterprise procurement, no public stock API.

    19 GPU models · 918 configurations

    Pricing not available

    $0.09/GB egress

    No public stock API

    Top GPUs: L4 · A10G · T4 · L40S

    Explore provider
  • Fluence logo

    Fluence

    Marketplace

    GPU marketplace of independent data centers; VMs and bare metal from named upstream clouds, priced in USDC.

    19 GPU models · 299 configurations · hourly billing

    $0.17 — $13.66/hr · 1× GPU

    Cheapest H100: $2.75/GPU-hr

    Egress unknown

    Live stock

    Top GPUs: A100-80GB · L40S · RTX A6000 · H100

    Explore provider
  • Akash Network logo

    Akash Network

    Marketplace

    Decentralized compute marketplace; network-average bid price per GPU and free GPU count from its public API.

    17 GPU models · 19 configurations

    $0.03 — $2.73/hr · 1× GPU

    Cheapest H100: $2.73/GPU-hr

    Egress unknown

    Live stock

    Top GPUs: RTX PRO 6000 · V100 · A100 · A100-80GB

    Explore provider
  • Lium logo

    Lium

    Marketplace

    Marketplace of independent GPU hosts priced per GPU, count and country; live stock from its public nodes feed.

    17 GPU models · 174 configurations

    $0.15 — $6.99/hr · 1× GPU

    Cheapest H100: $1.90/GPU-hr

    Egress unknown

    Live stock

    Top GPUs: RTX 5090 · RTX 3090 · RTX 4090 · RTX PRO 6000

    Explore provider
  • SaladCloud logo

    SaladCloud

    Marketplace

    Distributed network of consumer RTX GPUs with per-second billing, four priority tiers and hourly live stock.

    16 GPU models · 16 configurations · per-second billing

    $0.12 — $0.60/hr · 1× GPU

    Egress unknown

    Live stock

    Top GPUs: RTX 3080 · RTX 3080 Ti · RTX 3090 · RTX 3090 Ti

    Explore provider
  • OCI logo

    OCI

    Hyperscaler

    Oracle's hyperscaler cloud with bare-metal GPU shapes up to GB300 and MI355X; host capacity polled hourly.

    15 GPU models · 819 configurations

    $2.00 — $2.95/hr · 1× GPU

    $0.0085/GB egress

    Live stock

    Top GPUs: V100 · A10 · P100 · A100

    Explore provider
  • Verda logo

    Verda

    Specialist GPU cloud

    European GPU cloud with NVIDIA instances; prices and live per-location stock via its API.

    14 GPU models · 55 configurations

    $1.67 — $2.45/hr · 1× GPU

    Free egress

    Live stock

    Top GPUs: RTX PRO 6000 · H100 · H200 · A100-80GB

    Explore provider
  • AceCloud logo

    AceCloud

    Specialist GPU cloud

    GPU cloud with on-demand NVIDIA instances; full live stock coverage via its pricing API.

    12 GPU models · 46 configurations

    Pricing not available

    Egress unknown

    Live stock

    Top GPUs: RTX PRO 6000 · H200 · L4 · L40S

    Explore provider
  • Seeweb logo

    Seeweb

    Regional cloud

    Italian cloud provider with GPU servers in EU data centres; catalog pricing, no public stock API.

    11 GPU models · 43 configurations

    Pricing not available

    Egress unknown

    No public stock API

    Top GPUs: L40S · L4 · RTX A6000 · RTX PRO 6000 96GB

    Explore provider
  • Azure logo

    Azure

    Hyperscaler

    Microsoft's hyperscaler cloud with enterprise GPU VMs across many regions; no public stock API.

    10 GPU models · 571 configurations

    Pricing not available

    $0.087/GB egress

    No public stock API

    Top GPUs: T4 · A10 · A100-80GB · RTX PRO 6000

    Explore provider
  • Digital Ocean logo

    Digital Ocean

    Specialist GPU cloud

    Developer-friendly cloud with GPU Droplets (H100, H200, MI300X and more); prices and live stock via its API.

    9 GPU models · 16 configurations

    Pricing not available

    $0.01/GB egress

    Live stock

    Top GPUs: H100 · B300 · H200 · MI325X

    Explore provider
  • Lambda logo

    Lambda

    Specialist GPU cloud

    Specialist AI cloud for NVIDIA training GPUs; per-minute billing and region-aware live stock.

    9 GPU models · 51 configurations · per-minute billing

    $1.09 — $2.59/hr · 1× GPU

    Cheapest H100: $4.09/GPU-hr

    Free egress

    Live stock

    Top GPUs: H100 · A100 · RTX A6000 · A100-80GB

    Explore provider
  • PrimeIntellect logo

    PrimeIntellect

    Marketplace

    Compute marketplace aggregating GPU capacity across clouds; per-minute billing and live stock.

    9 GPU models · 43 configurations · per-minute billing

    $1.29 — $2.45/hr · 1× GPU

    Cheapest H100: $3.00/GPU-hr

    Free egress

    Live stock

    Top GPUs: H100 · RTX PRO 6000 · A100 · H200

    Explore provider
  • Hyperstack logo

    Hyperstack

    Specialist GPU cloud

    Specialist GPU cloud (NexGen Cloud) with per-minute billing and live stock for recent NVIDIA accelerators.

    8 GPU models · 34 configurations · per-minute billing

    $0.15 — $1.00/hr · 1× GPU

    Free egress

    Live stock

    Top GPUs: H100 · A100-80GB · RTX A4000 · L40

    Explore provider
  • Vultr logo

    Vultr

    Specialist GPU cloud

    Global cloud with GPU instances and bare metal across many regions; live stock via its API.

    8 GPU models · 16 configurations

    Pricing not available

    Egress unknown

    Live stock

    Top GPUs: L40S · A100-80GB · GH200 · B200

    Explore provider
  • Koyeb logo

    Koyeb

    Specialist GPU cloud

    Serverless platform with per-second GPU instances in 1-4× NVIDIA configs; capacity level from its public API.

    7 GPU models · 15 configurations · per-second billing

    $0.50 — $3.00/hr · 1× GPU

    Cheapest H100: $2.50/GPU-hr

    Egress unknown

    Live stock

    Top GPUs: A100-80GB · H100 · H200 · L40S

    Explore provider
  • Nebius logo

    Nebius

    Specialist GPU cloud

    Specialist AI infrastructure cloud with per-second billing and region-aware live stock.

    6 GPU models · 38 configurations · per-second billing

    $1.55 — $9.50/hr · 1× GPU

    Cheapest H100: $4.50/GPU-hr

    Free egress

    Live stock

    Top GPUs: L40S · B300 · H200 · RTX PRO 6000

    Explore provider
  • OVHcloud logo

    OVHcloud

    Regional cloud

    European cloud with GPU instances in EU and North American regions; live per-location stock.

    5 GPU models · 54 configurations

    $0.77 — $1.80/hr · 1× GPU

    Free egress

    Live stock

    Top GPUs: V100 · A100-80GB · H100 · L4

    Explore provider
  • Packet logo

    Packet

    Specialist GPU cloud

    GPU cloud with dedicated full-card 1-8× configs from its published SKU feed; stock status inline.

    4 GPU models · 14 configurations

    Pricing not available

    Egress unknown

    Live stock

    Top GPUs: A100-80GB · B200 · L40S · RTX 4090

    Explore provider
  • Gcore logo

    Gcore

    Specialist GPU cloud

    Global edge cloud with GPU bare-metal servers across its AI regions; stock arrives with each pricing run.

    3 GPU models · 12 configurations

    Pricing not available

    Egress unknown

    Live stock

    Top GPUs: H100 · A100 · L40S

    Explore provider
  • Hyperbolic logo

    Hyperbolic

    Marketplace

    GPU rentals with public on-demand pricing; availability is unknown when the API omits stock.

    3 GPU models · 16 configurations

    Pricing not available

    Free egress

    Live stock

    Top GPUs: H100 · H200 · RTX PRO 6000

    Explore provider
  • Yotta logo

    Yotta

    Regional cloud

    Indian cloud provider with NVIDIA GPU instances; catalog pricing, no public stock API.

    3 GPU models · 32 configurations

    Pricing not available

    Egress unknown

    No public stock API

    Top GPUs: RTX 5090 · H200 · RTX PRO 6000

    Explore provider
  • Hinode logo

    Hinode

    Specialist GPU cloud

    Dedicated L4, L40S and RTX PRO 6000 machines in US East and Frankfurt.

    2 GPU models · 10 configurations · per-minute billing

    Pricing not available

    $0.1/GB egress

    No public stock API

    Top GPUs: L40S · RTX PRO 6000

    Explore provider
  • Meshive GPU Cloud logo

    Meshive GPU Cloud

    Specialist GPU cloud

    Seoul GPU cloud running its own RTX 5090 and RTX PRO pods, billed per minute; live GPU counts from its API.

    2 GPU models · 9 configurations · per-minute billing

    $0.71 — $0.83/hr · 1× GPU

    Egress unknown

    Live stock

    Top GPUs: RTX 5090 · RTX PRO 5000

    Explore provider
  • Samsung Cloud Platform logo

    Samsung SDS's Korean cloud with GPU virtual servers; catalog pricing, no public stock API.

    2 GPU models · 12 configurations

    Pricing not available

    Egress unknown

    No public stock API

    Top GPUs: V100 · A100-80GB

    Explore provider
  • EcoHash logo

    EcoHash

    Specialist GPU cloud

    GPU cloud with RTX PRO 6000 and other workstation-class cards at hourly rates; live stock via API.

    1 GPU model · 4 configurations

    $1.89/hr · 1× GPU

    Egress unknown

    Live stock

    Top GPUs: RTX PRO 6000

    Explore provider
  • Fluidstack logo

    Fluidstack

    Specialist GPU cloud

    Specialist AI cloud for H100/H200 clusters; catalog pricing only — no open signup, so stock is blocked.

    1 GPU model · 1 configuration

    Pricing not available

    Free egress

    Stock blocked upstream

    Top GPUs: H100

    Explore provider
  • Hot Aisle logo

    Hot Aisle

    Specialist GPU cloud

    AMD Instinct bare-metal cloud (MI300X) with per-minute billing and live stock via its API.

    1 GPU model · 2 configurations · per-minute billing

    Pricing not available

    Free egress

    Live stock

    Top GPUs: MI300X

    Explore provider
  • IBM Cloud logo

    IBM Cloud

    Hyperscaler

    IBM's enterprise cloud with GPU virtual servers; catalog pricing, no public stock API.

    1 GPU model · 9 configurations

    Pricing not available

    $0.09/GB egress

    No public stock API

    Top GPUs: V100

    Explore provider

GPU cloud providers: frequently asked questions

Which GPU cloud providers does GPU Finder compare?

36 GPU cloud providers: the hyperscalers (AWS, Google Cloud, Azure, OCI, IBM Cloud), specialist AI clouds (Lambda, Nebius, Hyperstack, Verda, Yotta, Seeweb, Gcore, OVHcloud, Scaleway, Digital Ocean, Vultr, Koyeb, Samsung Cloud Platform), marketplaces (Vast.ai, Runpod, Shadeform, Lium, PrimeIntellect, Theta EdgeCloud, EcoHash, Akash) and bare-metal specialists (Hot Aisle, Packet, AceCloud, Fluidstack, Hyperbolic). Every provider page lists each GPU instance with its on-demand and spot price.

Which GPU cloud provider is the cheapest?

It depends on the GPU. Marketplaces such as Vast.ai and Runpod usually hold the floor on consumer cards (RTX 4090, RTX 5090) and often on H100, while specialist clouds like Lambda, Nebius and Hyperstack are competitive on multi-GPU H100, H200 and B200 nodes with no egress fees. The cheapest cloud GPU table ranks every model by its cheapest in-stock price per GPU-hour across all providers.

Which GPU cloud providers show live availability?

27 of the 36 providers expose a stock signal GPU Finder checks hourly. Their cards show a 7-day in-stock history and each instance carries a live badge plus a 30-day reliability score; hyperscalers without a public stock API show price only.

Which GPU cloud has the widest range of NVIDIA models?

In GPU Finder's current tracked catalog, Vast.ai has the widest NVIDIA range with 57 distinct models. Counts use canonical model names (including separately tracked memory variants), not instance sizes or GPU counts. Listings do not guarantee stock. Compare GPU models.

Which cloud providers list NVIDIA H100 GPUs?

H100 is listed by AceCloud, Akash Network, AWS, Azure, Digital Ocean, EmpirioLabs AI, Fluence, Fluidstack, Gcore, Google Cloud (GCP), Hyperbolic, Hyperstack, Koyeb, Lambda, Lium, Nebius, OCI, OVHcloud, PrimeIntellect, Runpod, Seeweb, Shadeform, Theta EdgeCloud, Vast.ai, Verda, Vultr in GPU Finder's current tracked catalog. Check individual listings for region, price, GPU count and current stock. Compare H100 listings.

Which cloud providers list NVIDIA H200 GPUs?

H200 is listed by AceCloud, AWS, Azure, Digital Ocean, EmpirioLabs AI, Fluence, Google Cloud (GCP), Hyperbolic, Hyperstack, Koyeb, Lium, Nebius, OCI, PrimeIntellect, Runpod, Seeweb, Shadeform, Theta EdgeCloud, Vast.ai, Verda, Yotta in GPU Finder's current tracked catalog. Check individual listings for region, price, GPU count and current stock. Compare H200 listings.

Which cloud providers list NVIDIA B200 GPUs?

B200 is listed by AWS, EmpirioLabs AI, Fluence, Google Cloud (GCP), Hyperstack, Lambda, Lium, Nebius, OCI, Packet, Runpod, Shadeform, Vast.ai, Verda, Vultr in GPU Finder's current tracked catalog. Check individual listings for region, price, GPU count and current stock. Compare B200 listings.

Which cloud providers list NVIDIA B300 GPUs?

B300 is listed by Akash Network, AWS, Digital Ocean, EmpirioLabs AI, Fluence, Lium, Nebius, OCI, Runpod, Shadeform, Verda in GPU Finder's current tracked catalog. Check individual listings for region, price, GPU count and current stock. Compare B300 listings.

Which GPU cloud providers have European regions?

AWS, Azure, Digital Ocean, Fluence, Fluidstack, Gcore, Google Cloud (GCP), Hinode, Hyperbolic, Hyperstack, IBM Cloud, Lambda, Lium, Nebius, OCI, Packet, PrimeIntellect, Seeweb, Shadeform, Vast.ai, Verda have tracked listings with explicitly European regions. Global or unspecified pools are excluded; a European listing does not guarantee stock or EU-only data residency. Confirm the exact location and terms before renting. Compare H100 in Europe. Compare H200 in Europe. Compare A100 in Europe. Compare A100-80GB in Europe.

Which GPU cloud providers bill per second or per minute?

per-second: EmpirioLabs AI, Koyeb, Nebius, Runpod, SaladCloud, Vast.ai. per-minute: Hinode, Hot Aisle, Hyperstack, Lambda, Meshive GPU Cloud, PrimeIntellect. per-hour: Fluence. These are recorded billing increments, not a promise of no minimum charge. Minimums and product-specific terms may apply; unrecorded billing is unknown. Estimate costs with billing increments.

Want the lowest price per GPU-hour by model instead? See the cheapest cloud GPU table, or a provider head-to-head.