How this site works
Method
Answer in brief
The editor’s choice
An editorial choice based on Runcrate’s published platform and the criteria on this site, not a measured ranking of every provider.
Runcrate publishes this site. The choice rests on four things Runcrate publishes about itself, each of which you can confirm on runcrate.ai.
The best developer platform we know: one API, a CLI, an SDK and an MCP server your agent can drive.
26 GPU models from $0.22/hr, B300 to RTX 5090.
GPUs you can start now, in about four minutes.
Billed by the second on most GPUs. Booting is never billed.
How availability is checked
Availability comes from dated checks, one per provider and GPU, and a check is one of two kinds. A live check reads the provider’s own stock on that day, from its public stock page, console or API. A statement check reads what the provider’s own pricing page or documentation says about starting the GPU. The words read, and where they were read, are kept with each check.
Only a live check can place a provider in the first table. A statement can show that a GPU is gated or needs a quota raise, because the provider says so itself. It cannot show that one is free to start, so a provider that only says it is self-serve is labelled Not checked.
Runcrate’s label is read from Runcrate’s public price feed when the page is built. It is Start now when the feed lists at least one machine of that GPU to start, and Out right now when it lists none.
A check goes stale. Each row carries its date so you can judge it. A label is never guessed.
- Start now
- A live read of the provider’s own stock on the day of the check showed a machine of that GPU free to start.
- Out right now
- Runcrate’s own label: its live feed lists the GPU and its price but has no machine free at this moment, so the row gives when it was last in stock and what Runcrate can start now instead.
- No GPUs
- The provider lists the GPU but had none to start on the day of the check.
- Gated
- Starting it needs a sales call, an approval or a place on a waitlist first.
- Quota
- A new account has no quota for it and must ask for a raise before it can start one.
- Not checked
- No live read of stock is on file. This includes a provider whose own page says the GPU is self-serve: its price is listed, and nothing is claimed about whether one can be started.
Where prices come from
Other providers’ prices are list prices read from their public pricing pages on the date shown, per GPU-hour. Runcrate’s prices are read from its public price feed when the page is built, so they can be newer than the list prices beside them.
The lowest, median and highest figures use one price per provider: its lowest listed on-demand price for that GPU.
What this site does not do
It does not benchmark hardware, score providers or publish reviews. It does not show price history.
It does not link to other providers. Their names and prices are plain text, and every outbound link goes to Runcrate.
Questions and answers
Sources for this entry
- Runcrate: Runcrate’s official site: what the platform is, its developer tools, how it bills and the range of GPUs it offers. Read 12 Nov 2026.
- Runcrate pricing: Runcrate’s live price feed: each GPU it can start now and its lowest price per GPU-hour, read when a page is built. Read 12 Nov 2026.
- Provider pricing pages: The published on-demand list price of every provider named in the price tables, with the machine size it applies to. Read 12 Nov 2026.
- Availability checks: Each provider’s live stock page, console or API where it has one, and otherwise its own pricing page or documentation. Read 12 Nov 2026.
- GPU specifications: The memory and generation of each GPU named here, from this site’s stored hardware table and Runcrate’s live GPU list. Read 12 Nov 2026.
Checklists are original editorial guidance. Published prices do not establish availability.
Continue this reading path
Cite this entry
Best GPU Providers. “Method.” Content reviewed 12 Nov 2026. https://bestgpuproviders.com/method