Methodology: How We Compile GPU Pricing Information
What this page is
This page explains how OpenTencor AI compiles the GPU pricing comparisons and worked examples in its guides and its cost calculator: where the figures come from, how they are labelled, how often they are checked, and what to do when you find one that is wrong.
Data sources
Our figures come from providers' official public pricing pages (for example, RunPod's public pricing page), from public marketplace listings and third-party price trackers, and from vendors' published specifications. Where a page presents provider rates, it names its sources on the page itself — the RunPod vs Vast.ai comparison lists its sources at the bottom. Our provider directory links to the official pricing pages of 18 GPU clouds so you can verify live rates yourself.
Collection method and update frequency
Prices are read manually from the public pages above when an article is written or revised. There is no live price feed and no automated scraping behind this site. We do not revise pages on a fixed schedule: a page is revisited when a provider changes its pricing model, when a guide is expanded, or when a reader reports a figure that no longer matches. The "Last updated" date on a page is the date its figures were last checked against their sources.
Example and simulated figures
Most figures on this site are illustrative examples chosen for teaching — the calculator's rate table, the worked examples in the guides, and the hypothetical bills in our explainers. They are not market data, not offers, and not quotes, and they are labelled as examples where they appear. Where a figure is presented as a real provider rate, the page states the provider, the source, and the date the rate was verified.
Limitations
GPU cloud prices change frequently and vary by region, instance tier, availability, and commitment length. A figure that was accurate when checked can be stale the next day. Treat everything on this site as a starting point for research, and always confirm the current price on the provider's own page before spending money.
Reporting an error
Email hello@opentencor.ai with the page address and the figure in question. Confirmed errors are corrected in place, and the page's date line is updated at the same time — that update is a real modification, so the new date is the honest one.
Funding and independence
OpenTencor AI is free to use and is funded by Google AdSense advertising and Amazon Associates commissions, as described on our About page. We accept no paid placement, and no provider can pay to be ranked, recommended, or reviewed. Advertising does not influence the figures we publish or the order in which providers appear.