V100 rents for $0.19 per GPU-hour today (neo-cloud on-demand median of 6 quotes, $0.19 to $0.79, 2026-09-27). 6 providers quote it on the latest day. Daily fixings per segment, 90-day history, capacity stress and the methodology behind every number.
| Provider | Market | Lowest USD/GPU-hr | Quotes |
|---|---|---|---|
| DataCrunch | Spot | $0.10 | 4 |
| DataCrunch | On-demand | $0.19 | 4 |
| RunPod | Community cloud | $0.19 | 2 |
| RunPod | Spot | $0.19 | 2 |
| Akash | Lease | $0.20 | 2 |
| Shadeform | On-demand | $0.39 | 9 |
| Lambda | On-demand | $0.79 | 2 |
| io.net | On-demand | $2.46 | 3 |
Lowest listed price per provider and market in the latest snapshot. The API returns every quote with its source URL; fixings are medians of eligible quotes, computed as described in the methodology. Prices are for renting the GPU by the hour, per GPU, before egress or storage.
Quote these fixings with the attribution "Source: corpusAI Cloud Pricing", a link to this page and the methodology version, under the data license. Providers are snapshotted several times a day. The page shows the latest day on which at least three providers reported quotes for any GPU; each segment's fixing is the median of that day's eligible quotes from the segment's members, taking each provider's latest snapshot of the day, so a fixing can rest on a single quote. The count is shown with every fixing.