01GPU cloud· manual survey 2026-10-14 · automated rows 2026-10-14
GPU cloud rental prices
What compute rents for. A venue-by-venue view of accelerator pricing with billing and evidence kept intact. Pick a chip — the statistics, history, and quotes below all follow it.
H100 · Cheapest on-demand
$1.99/GPU·hr
Voltage Park · n=12
H100 · Typical — median of 8 specialist clouds
$3.99/GPU·hr
Lowest $1.99 · highest $6.16
H100 · Big-cloud list price
$6.88/GPU·hr
AWS (p5) · n=4
H100 · venue spread
- Big-cloud list price ÷ Marketplace floor
- 3.2× AWS (p5) $6.88/GPU·hr ÷ Vast.ai $2.14/GPU·hr
- Typical ÷ Marketplace floor
- 1.9× $3.99/GPU·hr, median of 8 specialist clouds (Voltage Park $1.99 · Hyperstack $3.20 · Crusoe $3.90 · RunPod $3.99 · Lambda $3.99 · Together AI $3.99 · Nebius $4.50 · CoreWeave $6.16) ÷ Vast.ai $2.14/GPU·hr (RunPod and Voltage Park last observed Oct 11, 2026)
Different venues sell different things — SLAs, egress, and commitment terms — so the gap measures venue dispersion, not an arbitrage.
Comparable price history
Loading evidence ledger…
Provider dispersion
H100 · comparable provider-list range
On-demand list prices as each provider publishes them — the comparable set behind “Cheapest on-demand”. The dashed anchor is Kalshi-implied, a non-executable market reference rather than a provider quote: it settles on Ornn’s executed-trade index — not comparable with posted list prices.
- Azure (ND v5)On-demand$12.29 per GPU hour; Hyperscaler; On-demand; observed 2026-10-14
- Google Cloud (A3)On-demand$11.06 per GPU hour; Hyperscaler; On-demand; observed 2026-10-14
- Oracle OCIOn-demand$10.00 per GPU hour; Hyperscaler; On-demand; observed 2026-10-14
- AWS (p5)On-demand$6.88 per GPU hour; Hyperscaler; On-demand; observed 2026-10-14
- CoreWeaveOn-demand$6.16 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- NebiusOn-demand$4.50 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- RunPodOn-demand$3.99 per GPU hour; Specialist cloud; On-demand; observed 2026-10-11
- LambdaOn-demand$3.99 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- Together AIOn-demand$3.99 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- CrusoeOn-demand$3.90 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- HyperstackOn-demand$3.20 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14
- Voltage ParkOn-demand$1.99 per GPU hour; Specialist cloud; On-demand; observed 2026-10-11
Current published quotes
Loading the sortable table and expandable mobile evidence cards…
Kalshi-implied prices beside posted cloud prices
Weekly exact-time or explicitly labeled front-month monthly-average implied medians (Kalshi, OCT 16 2026 · 11:51 UTC), beside the range of on-demand list prices in each provider class and, apart, marketplace asks (manual survey 2026-10-14; automated rows through 2026-10-14). Capacity blocks, serverless, and secondary estimates are excluded. Anchors are Kalshi-implied medians, not current spot quotes; each settles on Ornn’s executed-trade index — not comparable with posted list prices. Moves show only between anchors that passed the ladder-quality gate.
| Chip | Implied anchor | Prior snapshot | 30D same contract | Front-month avg | Hyperscaler | Specialist | Marketplace |
|---|---|---|---|---|---|---|---|
| B200 price historyHBM3e · 180GB | $7.72Kalshi-implied · Oct avgthin · $7.62–7.82 | No reliable movethin anchor | No reliable move | $7.72thin · $7.62–7.82 | $14.00–$14.241 | $6.00–$8.602 | $6.253 |
| H200 price historyHBM3e · 141GB | $5.98Kalshi-implied · Oct avgquality-passed · $5.63–6.14 | No reliable moveprior anchor thin | +9.1%5 closes · 10-11–10-16 | $5.98quality-passed · $5.63–6.14 | $7.91–$10.604 | $3.99–$6.315 | $3.686 |
| H100 price historySXM · 80GB | $2.87Kalshi-implied · Oct avgquality-passed · $2.73–3.00 | −0.7%14h 8m | +2.1%6 closes · 10-11–10-16 | $2.87quality-passed · $2.73–3.00 | $6.88–$12.297 | $1.99–$6.168 | $2.149 |
| A100 price historySXM4 · 80GB | $1.09Kalshi-implied · Oct avgthin · $0.87–1.31 | No reliable movethin anchor | No reliable move | $1.09thin · $0.87–1.31 | $3.43–$5.0710 | $1.35–$2.7911 | — |
| RTX 5090 price historyGDDR7 · 32GB | $0.66Kalshi-implied · Oct avgthin · $0.62–0.70 | No reliable movethin anchor | No reliable move | $0.66thin · $0.62–0.70 | — | $1.1912 | $0.25–$0.6013 |
Provider quote sources13 notes
- 1 Oracle OCI $14.00 · AWS (p6) $14.24
- 2 Hyperstack $6.00 · Lambda $6.69 · RunPod $7.99 · Together AI $8.19 · Nebius $8.50 · CoreWeave $8.60
- 3 Vast.ai $6.25
- 4 AWS (p5en) $7.91 · Oracle OCI $10.00 · Azure (ND v5) $10.60
- 5 Hyperstack $3.99 · Crusoe $4.29 · RunPod $5.29 · Nebius $5.40 · Together AI $5.99 · CoreWeave $6.31
- 6 Vast.ai $3.68
- 7 AWS (p5) $6.88 · Oracle OCI $10.00 · Google Cloud (A3) $11.06 · Azure (ND v5) $12.29
- 8 Voltage Park $1.99 · Hyperstack $3.20 · Crusoe $3.90 · RunPod $3.99 · Lambda $3.99 · Together AI $3.99 · Nebius $4.50 · CoreWeave $6.16
- 9 Vast.ai $2.14
- 10 AWS (p4de) $3.43 · Oracle OCI $4.00 · Google Cloud (A2 Ultra) $5.07
- 11 Hyperstack $1.35 · RunPod $1.79 · Crusoe $2.30 · CoreWeave $2.70 · Lambda $2.79
- 12 RunPod $1.19
- 13 Salad $0.25 · Vast.ai $0.60
Marketplace tape · intraday ask floor — listed offers, not clearing prices
Vast.ai verified on-demand asks per chip, shown over the last 48h. 152 captures across 30 days (hourly schedule · ~5 captures/day), through 2026-10-16 11:56 UTC. Missed hours are gaps, never interpolated.
- H100$2.16floormedian $7.30 · 6 offers · 5/48h observed
- H200$5.00floormedian $5.00 · 6 offers · 2/48h observed
- B200$6.25floormedian $8.81 · 8 offers · 7/48h observed
- RTX 4090$0.40floormedian $0.79 · 43 offers · 13/48h observed
- RTX 5090$0.48floormedian $1.12 · 53 offers · 13/48h observed
Ask floor — listed offers, not clearing prices. Full rolled archive: /marketplace-tape.json.