Category guide
Compute Price Watch: category guide
Answer in brief
A list price is not a price you can get. Before you compare compute prices, check three things: that you can start the GPU today, how many you must rent at once, and what unit you are billed in.
A list price is not a price you can get
Most price pages for the newest GPUs describe an offer, not stock. The card is sold out, or it is held for customers who talk to sales first, or a new account has no quota for it and must ask. The number on the page stays the same in all three cases.
So a compute price is worth watching only together with whether it can be started. That is what this page records beside every price, with the day it was checked.
What a compute price has to tell you
A usable compute price has four parts: the figure, the unit it is billed in, the smallest amount you can rent, and whether it can be started today. Most comparisons give only the first.
Read every price per GPU-hour, so a single card and an 8-GPU node can sit in one column. Then read the smallest rental, because that decides what an hour really costs you. On 12 Nov 2026, the lowest listed hourly total for an 8-GPU B300 machine was $52.80.
Before you trust a compute price
Work through the questions in this order. Each one removes options the next would have wasted time on.
- Can you start it today?
Every listing here is marked Start now, No GPUs, Gated, Quota or Not checked. A price you cannot start is not a price.
- What is the smallest thing you can rent?
Some B300 and B200 prices are per GPU inside an 8-GPU node. You pay for all eight.
- What unit are you billed in?
By the second, the minute or the hour. Ask too whether the minutes a machine takes to boot are billed.
- Is it on demand?
A spot machine can be taken back, and a reserved price needs a term paid ahead. Neither compares with an on-demand price.
- What else is on the bill?
Storage, data transfer and idle time. None of them is in the GPU-hour.
- When was it read?
Prices move within a day. Read the date beside the price before you rely on it.
Terms this guide uses
- On-demand
An on-demand price is what you pay to start a machine when you want it and stop it when you are done, with no term and no risk of losing it.
- Spot
A spot price is a lower price for spare capacity that the provider can take back at short notice.
- Reserved
A reserved price is a lower hourly rate in return for a commitment to a term, often paid ahead, and billed for every hour of that term.
- Node
A node is one physical machine with several GPUs in it, most often eight, rented as a whole.
- Quota
A quota is the number of machines of a given kind that your account is allowed to run, and for large GPUs a new account often starts at zero.
Where Runcrate fits
Compute Price Watch names one provider as its editor’s choice and gives the reasons in full, so you can weigh them against your own brief.
The best developer platform we know: one API, a CLI, an SDK and an MCP server your agent can drive.
Billed by the second on most GPUs. Booting is never billed.
26 GPU models from $0.23/hr, B300 to RTX 5090.
GPUs you can start now, in about four minutes.
Questions and answers
Sources for this entry
- Runcrate: Runcrate’s official site: what the platform is, how it bills and the range of GPUs it offers. Read 12 Nov 2026.
- Runcrate pricing: Runcrate’s live price feed: each GPU it lists, whether one can be started now, and the price it starts from 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.
Checklists are original editorial guidance. Published prices do not establish availability.
Continue this reading path
Cite this entry
Compute Price Watch. “Compute Price Watch: category guide.” Content reviewed 12 Nov 2026. https://computepricewatch.com/guide