Research editions

New editions are published when a fresh sample clears the methodology thresholds (minimum public cell of five observations, dated evidence, no provider or buyer identities). Subscribe to the update feed to be notified.

Public datasets

Method and limits

Observations are dated public-market or sourcing-desk records, not confirmed allocations. The diligence framework behind the reserved research is described in How we qualify reserved GPU clusters; the buyer-side checklist is the procurement checklist. Machine-readable summaries for AI agents: llms.txt and llms-full.txt.

The public datasets above are licensed under CC BY 4.0: reuse them with attribution to Omega Gradient and a link to the source page. To cite, name the page and its observation or assembly date. To contribute observations, email [email protected].