How the index works

What we measure

We collect published on-demand list prices for cloud GPU rentals and normalize every listing to USD per GPU per hour. The headline figure for each listing is total_price_usd_per_hour — the instance total divided by GPU count. This is always explainable.

Three prices, never invented

• Total $/GPU/hr — always present.
• Published GPU component — only when a provider explicitly splits out the GPU price.
• Estimated GPU component — only when we estimate it, always labelled EST.

When a bundled instance (e.g. a hyperscaler node) can't be separated into a GPU component, we show not separable rather than a fabricated number.

Quality tiers

Each listing is tagged Reliable, Community, or Experimental. Only index-eligible listings enter the median. A listing is excluded with an explicit reason — community/experimental tier, unknown form factor, aggregator source, suspicious validation, non-comparable serverless, or staleness over 12 hours.

Freshness & staleness

• < 2h — fresh, in the median.
• 2–12h — stale warning, still in the median.
• 12–24h — excluded from the median, still shown.
• > 24h — hidden by default.

History

Historical curves are built from immutable daily snapshots, each carrying a methodology version and content hash. Until snapshots accumulate, a SKU shows “Tracking starts from launch date” — we never plot placeholder data.

Boundaries

This is an informational index of list prices, not a settlement-grade transaction benchmark. List price ≠ transaction price. Verify with the provider before provisioning. Snapshots are licensed CC BY 4.0.