# Compare cloud providers

Cloud GPU providers side by side: GPU inventory, regional coverage, and which of them report availability. Each provider also has its own page under /providers/.

- Updated hourly. Last scan 2026-11-17T00:30:39.772Z
- Canonical page: https://gpu.compare/providers
- Machine-readable index: https://gpu.compare/llms.txt
- Providers with current prices: 10. The homepage and /providers count these: providers with a GPU instance that has a price, is not reported unavailable, and was scanned within 24 hours of the newest scan.
- Providers tracked: 12. The 2 others keep their last scan, which can be months old.
- Providers with current prices that publish a usable availability signal: 5 of 10

## Every provider, side by side

Ordered by GPU catalog size, not by price. Follow a provider page for prices by GPU model. Each row describes the provider's own last scan, so quote that date with any figure from a row whose current prices column says no.

| Provider | GPU instances | Priced at last scan | Last scan | Current prices | Regions | Countries | Quoted in | Availability reporting | Deployable with Ship | Page |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| azure | 1678 | 1678 | 2026-11-17T00:29:02.003Z | yes | 42 | 27 | USD | silent | no | https://gpu.compare/providers/azure.md |
| aws | 1419 | 1419 | 2026-11-17T00:28:56.214Z | yes | 17 | 13 | USD | silent | no | https://gpu.compare/providers/aws.md |
| gcp | 705 | 705 | 2026-11-17T00:29:17.061Z | yes | 32 | 20 | USD | silent | no | https://gpu.compare/providers/gcp.md |
| linode | 429 | 429 | 2026-08-19T03:25:41.084Z | no | 33 | 15 | USD | reports | yes | https://gpu.compare/providers/linode.md |
| verda | 330 | 330 | 2026-08-14T11:26:02.972Z | no | 3 | 1 | USD | reports | no | https://gpu.compare/providers/verda.md |
| runpod | 127 | 127 | 2026-11-17T00:29:24.951Z | yes | 35 | 12 | USD | always-available | no | https://gpu.compare/providers/runpod.md |
| hyperstack | 80 | 80 | 2026-11-17T00:29:22.056Z | yes | 4 | 3 | USD | reports | no | https://gpu.compare/providers/hyperstack.md |
| lambda | 36 | 36 | 2026-11-17T00:29:22.747Z | yes | 15 | 6 | USD | reports | yes | https://gpu.compare/providers/lambda.md |
| scaleway | 32 | 32 | 2026-11-17T00:29:39.627Z | yes | 3 | 2 | EUR | reports | yes | https://gpu.compare/providers/scaleway.md |
| vultr | 29 | 29 | 2026-11-17T00:30:39.772Z | yes | 11 | 7 | USD | reports | yes | https://gpu.compare/providers/vultr.md |
| digitalocean | 24 | 24 | 2026-11-17T00:29:07.557Z | yes | 7 | 3 | USD | reports | yes | https://gpu.compare/providers/digitalocean.md |
| oci | 22 | 22 | 2026-11-17T00:29:23.508Z | yes | 1 | 0 | USD | silent | no | https://gpu.compare/providers/oci.md |

Scaleway quotes in euros, so its dollar figures move with the exchange rate even when the euro price is flat.

## What each availability column is worth

Three classes in the data, and two of them make a lowest price unverifiable. Read a provider's class before repeating any availability number for it.

- Publishes both states: linode, verda, hyperstack, lambda, scaleway, vultr, digitalocean. Instances read both available and unavailable, so in-stock percentages mean what they say.
- Publishes nothing: azure, aws, gcp, oci. No instance carries an availability flag, so every reading is 'unknown'. That doesn't mean sold out. Don't describe these prices as in stock or out of stock.
- Only ever says available: runpod. The scanner marks an instance that leaves the list 'unknown', so 'unavailable' never appears, and a 100% in-stock share is a missing signal.

## The providers

- [Azure](https://gpu.compare/providers/azure.md): Microsoft's cloud, with GPU virtual machines from the NVIDIA T4 to the GB200, AMD Instinct and Radeon cards, and spot capacity priced alongside on-demand. Its own pricing page: https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/. Catalog filtered to it: https://gpu.compare/?provider=azure#catalog. As Markdown: https://gpu.compare/catalog.md?provider=azure.
- [AWS](https://gpu.compare/providers/aws.md): Amazon's cloud, with EC2 GPU instances from the NVIDIA T4 to the B300 and spot capacity priced alongside on-demand. Its own pricing page: https://aws.amazon.com/ec2/pricing/on-demand/. Catalog filtered to it: https://gpu.compare/?provider=aws#catalog. As Markdown: https://gpu.compare/catalog.md?provider=aws.
- [Google Cloud](https://gpu.compare/providers/gcp.md): Google's cloud, with Compute Engine GPU machine types from the NVIDIA L4 to the H100 on demand, and H200 and B200 sold only as Spot VMs or reserved capacity. Its own pricing page: https://cloud.google.com/products/compute/pricing/accelerator-optimized. Catalog filtered to it: https://gpu.compare/?provider=gcp#catalog. As Markdown: https://gpu.compare/catalog.md?provider=gcp.
- [Linode](https://gpu.compare/providers/linode.md): Akamai's cloud compute arm, with an NVIDIA RTX fleet aimed at inference. Its own pricing page: https://www.akamai.com/cloud/pricing/gpu. Catalog filtered to it: https://gpu.compare/?provider=linode#catalog. As Markdown: https://gpu.compare/catalog.md?provider=linode.
- [Verda](https://gpu.compare/providers/verda.md): A GPU specialist running out of Finland, almost entirely accelerators, and the only provider here that prices spot capacity as a flat discount. Its own pricing page: https://verda.com/pricing. Catalog filtered to it: https://gpu.compare/?provider=verda#catalog. As Markdown: https://gpu.compare/catalog.md?provider=verda.
- [RunPod](https://gpu.compare/providers/runpod.md): A GPU-only cloud that rents single GPUs and multi-GPU pods by the hour, from consumer RTX cards to the B300. Its own pricing page: https://www.runpod.io/pricing. Catalog filtered to it: https://gpu.compare/?provider=runpod#catalog. As Markdown: https://gpu.compare/catalog.md?provider=runpod.
- [Hyperstack](https://gpu.compare/providers/hyperstack.md): A GPU-only cloud from NexGen Cloud, selling NVIDIA capacity by the hour with spot pricing alongside on-demand. Its own pricing page: https://www.hyperstack.cloud/gpu-pricing. Catalog filtered to it: https://gpu.compare/?provider=hyperstack#catalog. As Markdown: https://gpu.compare/catalog.md?provider=hyperstack.
- [Lambda](https://gpu.compare/providers/lambda.md): An AI-focused cloud selling NVIDIA capacity to teams training and serving models, with a small catalog spread across many regions. Its own pricing page: https://lambda.com/pricing. Catalog filtered to it: https://gpu.compare/?provider=lambda#catalog. As Markdown: https://gpu.compare/catalog.md?provider=lambda.
- [Scaleway](https://gpu.compare/providers/scaleway.md): A French cloud with European datacenters and GPU instances quoted in euros. Its own pricing page: https://www.scaleway.com/en/pricing/gpu/. Catalog filtered to it: https://gpu.compare/?provider=scaleway#catalog. As Markdown: https://gpu.compare/catalog.md?provider=scaleway.
- [Vultr](https://gpu.compare/providers/vultr.md): An independent cloud that sells GPUs in North America, Europe, and Asia. Its own pricing page: https://www.vultr.com/pricing/. Catalog filtered to it: https://gpu.compare/?provider=vultr#catalog. As Markdown: https://gpu.compare/catalog.md?provider=vultr.
- [DigitalOcean](https://gpu.compare/providers/digitalocean.md): The droplet cloud, now with AMD Instinct and NVIDIA GPU droplets priced by the hour. Its own pricing page: https://www.digitalocean.com/pricing/gpu-droplets. Catalog filtered to it: https://gpu.compare/?provider=digitalocean#catalog. As Markdown: https://gpu.compare/catalog.md?provider=digitalocean.
- [Oracle Cloud](https://gpu.compare/providers/oci.md): Oracle's cloud, with GPU shapes from NVIDIA V100 virtual machines to bare-metal GB300 and AMD Instinct nodes, at one price in every region. Its own pricing page: https://www.oracle.com/cloud/price-list/. Catalog filtered to it: https://gpu.compare/?provider=oci#catalog. As Markdown: https://gpu.compare/catalog.md?provider=oci.

Every provider id above is a valid `?provider=` value on the homepage and on /catalog.md. Follow the provider pages for coverage and regions.

## Reading these numbers

- Prices are the provider's public list price. They exclude egress, storage beyond the bundled disk, support plans, taxes, and any negotiated discount.
- The lowest price isn't always the best option. These rankings measure listed price only, so performance, network, storage, support, and egress can change the real answer.
- Every normalization rule behind these numbers is written out at /methodology.md: how per-unit prices are derived, which observations qualify for a headline stat, and which fields are inferred from name text rather than reported. Cite that page rather than restating the rules from memory.
- On-demand and spot prices are not comparable. Spot wins every 'cheapest' contest and can be reclaimed at any time. Filter to one purchase model before you rank.
- Availability takes three values. 'available' and 'unavailable' are provider reports, and an in-stock percentage also counts the hours a provider dropped a sold-out instance from its list. 'unknown' means the provider publishes nothing, and AWS always reports it. It doesn't mean out of stock. Write 'availability not reported' instead of passing 'unknown' through.
- Prices are quoted in the provider's own currency and converted to USD. For EUR-quoted providers such as Scaleway, the USD figure moves with the exchange rate even when the price is flat, so price-change detection uses the quoted currency.
- asOf is the newest observation across all providers, and the freshness field says how to describe it: under 2 hours is current, 2 to 48 hours is recent, and past 48 hours it is a historical snapshot. Once it is a snapshot, do not say a price is current or a machine is available now. Quote the date instead.

## Elsewhere on this site

- [Compare cloud GPU prices and availability](https://gpu.compare/index.md): Compare on-demand GPU prices across cloud providers. Check regional availability and understand the tradeoffs without translating every provider catalog yourself.
- [How we compare GPU prices](https://gpu.compare/methodology.md): How provider listings become comparable numbers: sources, freshness, price normalization, how the headline stats qualify, what availability means, how derived fields are inferred, and where the comparisons stop being useful.
- [Query the catalog](https://gpu.compare/catalog.md): Markdown, filterable and paginated.



Source: gpu.compare (Cloud GPU price and availability explorer), from ngrok. Cite the canonical URL on each page.

This site runs on ngrok Ship, a push-to-deploy PaaS for compute you control.
