01GPU Prices· manual survey 2026-10-20 · automated rows 2026-10-20

GPU rental prices

Choose a GPU to compare provider prices, billing terms and observation dates.

H100 rental prices

Posted prices · USD per GPU-hour

Select accelerator

H100 · current published quotes

Hourly prices retain each venue’s billing model, cheapest first. Observation dates vary by row: unavailable sources retain their last valid quote, never a new observation date. All rows: Provisional — Dated source-page observation; the raw payload is not yet pinned to the archive.

9 venues

Search providersBilling modelSort by

9 of 9 quotes

Your filters, sorting and columns are remembered on this device.

Current H100 prices per GPU-hour by provider and billing model
Provider	$ / GPU·hr	Billing	Observed	Details
Hyperstack — open source page	$3.20	On-demand	2026-10-14	Details Provider type Specialist Pricing notes On-demand table; reserved and spot rates are separate. Source Source page · Hyperstack published pricing ↗ Configuration H100 SXM 80GB; per GPU-hour Region / billing Provider regions vary · On-demand
Modal — open source page	$3.95	Serverless / per-second	2026-10-20	Details Provider type Specialist Pricing notes GPU-only base rate $0.001097/second × 3,600, for preemptible execution with no region pinned; non-preemptible execution costs 3× and region selection 1.15–1.75× the base price; CPU, memory, and other resources billed separately. Source Source page · Modal published pricing ↗ Configuration H100 SXM GPU task; one GPU; billed per second Region / billing Provider regions vary · Serverless / per-second
Lambda — open source page	$3.99	On-demand	2026-10-14	Details Provider type Specialist Pricing notes 8-GPU Instances tier; applicable sales tax/VAT/GST extra. Smaller instances and 1-Click Clusters have different rates. Source Source page · Lambda published pricing ↗ Configuration 8-GPU instance; H100 SXM 80GB; published per GPU-hour Region / billing Provider regions vary · On-demand
Together AI — open source page	$3.99	On-demand	2026-10-20	Details Provider type Specialist Pricing notes GPU Clusters hourly on-demand rate; dedicated inference and preemptible rates differ. Source Source page · Together AI published pricing ↗ Configuration H100 HGX GPU Cluster; per GPU-hour Region / billing Provider regions vary · On-demand
Nebius — open source page	$4.50	On-demand	2026-10-20	Details Provider type Specialist Pricing notes On-demand rate in effect since October 1, 2026 (pricing.md table 'Item — from October 8, 2026'). Preemptible capacity uses dynamic spot pricing from $0.79. Source Source page · Nebius compute pricing documentation ↗ Configuration NVIDIA H100 NVLink (eu-north1); per GPU-hour; preset of 16 vCPUs and 200GB RAM per GPU per nebius.com/prices Region / billing Provider regions vary · On-demand
AWS (p5) — open source page	$6.88	On-demand	2026-10-20	Details Provider type Hyperscaler Pricing notes Post Jun-2025 cut Source Source page · AWS Price List API ↗ Configuration Per GPU within p5.48xlarge (8× H100 SXM) Region / billing us-east-1 (US East, N. Virginia) · On-demand
Oracle OCI — open source page	$10.00	On-demand	2026-10-14	Details Provider type Hyperscaler Pricing notes Published USD per GPU-hour; multiply by eight for the full bare-metal instance. Source Source page · Oracle OCI published pricing ↗ Configuration BM.GPU.H100.8; 8 × H100 80GB Region / billing OCI public USD price list; regional availability varies · On-demand
Google Cloud (A3) — open source page	$11.06	On-demand	2026-10-20	Details Provider type Hyperscaler Pricing notes Published instance rate $88.49/hour divided by 8, rounded to four decimals; bundled CPU, memory and local SSD included. Source Source page · Google Cloud (A3) published pricing ↗ Configuration a3-highgpu-8g; 8 × H100, 208 vCPUs, 1,871GB memory, 6,000GiB local SSD Region / billing Google public reference table; regional availability varies · On-demand
Azure (ND v5) — open source page	$12.29	On-demand	2026-10-20	Details Provider type Hyperscaler Source Source page · Azure Retail Prices API ↗ Configuration Per GPU within ND96isr H100 v5 (8× H100) Region / billing westus3 (cheapest US region) · On-demand

H100 · Cheapest on-demand

$3.20/GPU·hr

Hyperstack · n=8

H100 · Typical — median of 4 specialist clouds

$3.99/GPU·hr

Lowest $3.20 · highest $4.50

H100 · Big-cloud list price

$6.88/GPU·hr

AWS (p5) · n=4

H100 · venue spread

No marketplace floor for H100 — both venue-spread ratios need one as their shared denominator.

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…

1W · 1M · 3M · 1Y · ALL

Provider dispersion

H100 · comparable provider-list range

On-demand list prices as each provider publishes them — the comparable set behind “Cheapest on-demand”. Values are US dollars per GPU-hour.

Specialist cloudHyperscaler

Azure (ND v5)On-demand$12.29 per GPU hour; Hyperscaler; On-demand; observed 2026-10-20
Google Cloud (A3)On-demand$11.06 per GPU hour; Hyperscaler; On-demand; observed 2026-10-20
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 2026-10-20
NebiusOn-demand$4.50 per GPU hour; Specialist cloud; On-demand; observed 2026-10-20
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 2026-10-20
HyperstackOn-demand$3.20 per GPU hour; Specialist cloud; On-demand; observed 2026-10-14

Kalshi-implied GPU prices

Estimates from prediction markets, not prices available to rent at. Market snapshot: OCT 21 2026 · 21:57 UTC.

Thin market: insufficient quoted depth for a reliable trend. Ranges show the bid/ask bracket; no reliable move is shown when either endpoint fails the quality gate.

Price basis and methodology

Weekly exact-time or explicitly labeled front-month monthly-average implied medians. Anchors are Kalshi-implied medians, not current spot quotes; each settles on Ornn’s executed-trade index — not comparable with posted list prices.

Market-implied GPU rental estimates and comparable changes; posted provider quotes are in a separate section
Chip	Implied anchor	Change vs prior snapshot	30D same contract	Front-month avg
B200 price historyHBM3e · 180GB	$8.24Kalshi-implied · Oct avgthin · $8.20–8.29Thin market · no reliable trend	No reliable movethin anchor	No reliable move	$8.24thin · $8.20–8.29
H200 price historyHBM3e · 141GB	$6.01Kalshi-implied · Oct avgthin · $5.83–6.20Thin market · no reliable trend	No reliable movethin anchor	No reliable move	$6.01thin · $5.83–6.20
H100 price historySXM · 80GB	$2.89Kalshi-implied · Oct avgthin · $2.83–2.96Thin market · no reliable trend	No reliable movethin anchor	No reliable move	$2.89thin · $2.83–2.96
A100 price historySXM4 · 80GB	$0.90Kalshi-implied · Oct avgthin · $0.87–0.93Thin market · no reliable trend	No reliable movethin anchor	No reliable move	$0.90thin · $0.87–0.93
RTX 5090 price historyGDDR7 · 32GB	$0.66Kalshi-implied · Oct avgthin · $0.62–0.71Thin market · no reliable trend	No reliable movethin anchor	No reliable move	$0.66thin · $0.62–0.71

Vast.ai marketplace prices were withdrawn on 2026-10-20: Vast's terms bar their use in price comparisons. Correction record.

CoreWeave prices were withdrawn on 2026-10-20: CoreWeave's terms bar displaying its site's content with services that are not its own. Correction record.

Crusoe prices were withdrawn on 2026-10-20: Crusoe's terms bar copying, publishing or displaying its site's content without written permission. Correction record.

Salad prices were withdrawn on 2026-10-20: Salad's terms bar using or displaying any part of its site without written consent. Correction record.

RunPod prices were withdrawn on 2026-10-20: RunPod's terms bar compiling its data into a collection or database without written permission. Correction record.

Voltage Park prices were withdrawn on 2026-10-20: Voltage Park's terms bar collecting data from its platform without written permission. Correction record.

GetDeploying estimates were withdrawn on 2026-10-20: the quote tables list providers' own prices only. Correction record.

AIMultiple's market-average estimate was withdrawn on 2026-10-20: the quote tables list providers' own prices only. Correction record.