Changelog
What changed in Intelaris AI Compute Prices and on this website, by date.
- Partner page with an application form for GPU providers. English is the main language of the website; German is chosen in the globe menu.
- Price area rebuilt: GPU cards by segment, the table as a second view, and a quick view with the index for each pricing model.
- GPU index defined: the median across direct providers of each provider's cheapest comparable on-demand price per GPU hour, from its latest collection in the past three days, without outliers, and only from three providers.
- The price engine runs continuously on servers in the EU, with daily price history and a daily encrypted backup.
- Insights with categories, search and article pages. Figures in articles come from the price data.
- Price updates: Verda H100, H200 and B200; RunPod H100 PCIe (Secure Cloud). New providers: Seeweb, OUTSCALE, STACKIT, Civo.
- Listing rule: cloud providers running GPUs in data centers, and hyperscalers. Marketplaces, decentralized networks and inference services are no longer listed.
- Status page with the state of the price data.
- GPU pages with specifications and the list of providers. «From» and median use listed on-demand offers.
- «Get this GPU» opens a request window prefilled with the offer.
- Website in English and German. Currencies USD, EUR and CHF, converted from the original price with dated ECB reference rates.
- First version of AI Compute Prices: filters for GPU, region and pricing model, one page per GPU and per country.
- First price collection from public price lists and pricing APIs, each price with provider, location, currency, date and source.