Home / Price board
GPU price board · as of Nov 20, 2026
On-demand GPU prices, sourced.
Published on-demand list prices in US dollars per GPU-hour for five popular data-center GPUs, read from each provider's public pricing page. Every row links to its source. Unknowns are marked, not guessed.
$3.94H100 median of 14 prices · low $1.99
$5.35H200 median of 10 prices · low $2.50
$8.50B200 median of 9 prices · low $3.50
$2.00A100 median of 7 prices · low $1.09
How to read this. Prices are list prices for on-demand use as published on Nov 20, 2026; multi-GPU node prices are divided by GPUs per node. "From" means the provider's lowest advertised price for that GPU family; "Range" is a marketplace's advertised band. Configurations differ (SXM vs PCIe/NVL, interconnect, CPU, RAM, storage), so the cheapest row is not always the best fit. Prices change often; check the source before you rely on one. We may earn referral fees from some providers listed; that does not change the data shown.
| Type | Configuration & notes | Source | |||
|---|---|---|---|---|---|
| H100 | Voltage Park | $1.99 | From Lowest listed | HGX H100 From price, no contract (pricing FAQ) |
Pricing page ↗ |
| H100 | GMI Cloud | $2.60 | From | H100 | Pricing page ↗ |
| H100 | Denvr Dataworks | $3.00 | List | H100 SXM, 8x, InfiniBand | Pricing page ↗ |
| H100 | Massed Compute | $3.14 | List | H100 SXM5, 1x H100 family from $2.73 |
Pricing page ↗ |
| H100 | Thunder Compute | $3.20 | List | H100 SXM | Pricing page ↗ |
| H100 | Jarvis Labs | $3.49 | List | H100 SXM Spot $1.54 |
Pricing page ↗ |
| H100 | Prime Intellect | $3.50 | List | H100 Hourly, no contract |
Pricing page ↗ |
| H100 | RunPod | $4.39 | List | H100 SXM, Pods H100 PCIe $2.89 |
Pricing page ↗ |
| H100 | Nebius | $4.50 | List | HGX H100 | Pricing page ↗ |
| H100 | Lambda | $4.59 | List | HGX H100, 8x instance 1x H100 SXM listed at $4.89 |
Pricing page ↗ |
| H100 | Together AI | $5.49 | List | HGX H100 Reserved via sales |
Pricing page ↗ |
| H100 | Crusoe | $5.50 | List | HGX H100 Listed as rising to $6.30 on Dec 1, 2026 |
Pricing page ↗ |
| H100 | Verda | $5.51 | List | H100 SXM5 Spot listed at half price |
Pricing page ↗ |
| H100 | CoreWeave | $6.16 | List | HGX H100, 8-GPU node $49.24 per 8-GPU node-hour |
Pricing page ↗ |
| H100 | Vast.ai | n/a | Not published | Live marketplace pricing; no fixed list | Pricing page ↗ |
| H200 | Hydra Host | $2.50–$3.20 | Range Lowest listed | H200, bare metal marketplace | Pricing page ↗ |
| H200 | GMI Cloud | $3.20 | From | H200 | Pricing page ↗ |
| H200 | Massed Compute | $3.62 | List | H200 NVL, 1x NVL (PCIe) form factor |
Pricing page ↗ |
| H200 | Jarvis Labs | $4.59 | List | H200 SXM Spot $2.29 |
Pricing page ↗ |
| H200 | RunPod | $5.29 | List | H200, Pods | Pricing page ↗ |
| H200 | Nebius | $5.40 | List | HGX H200 | Pricing page ↗ |
| H200 | Prime Intellect | $5.48 | List | H200 Hourly, no contract |
Pricing page ↗ |
| H200 | Together AI | $5.99 | List | HGX H200 Reserved via sales |
Pricing page ↗ |
| H200 | Crusoe | $6.00 | List | HGX H200 Listed as rising to $7.50 on Dec 1, 2026 |
Pricing page ↗ |
| H200 | CoreWeave | $6.31 | List | HGX H200, 8-GPU node $50.44 per 8-GPU node-hour |
Pricing page ↗ |
| H200 | Denvr Dataworks | n/a | Not published | Reserved only | Pricing page ↗ |
| H200 | Lambda | n/a | Not published | Not listed on the on-demand page | Pricing page ↗ |
| H200 | Vast.ai | n/a | Not published | Live marketplace pricing; no fixed list | Pricing page ↗ |
| H200 | Verda | n/a | Not published | Offered; price not captured | Pricing page ↗ |
| B200 | Hydra Host | $3.50–$5.00 | Range Lowest listed | B200, bare metal marketplace | Pricing page ↗ |
| B200 | GMI Cloud | $5.00 | From | B200 Listed as limited availability |
Pricing page ↗ |
| B200 | Lambda | $7.69 | List | HGX B200, 8x instance | Pricing page ↗ |
| B200 | RunPod | $7.99 | List | B200, Pods | Pricing page ↗ |
| B200 | Nebius | $8.50 | List | HGX B200 | Pricing page ↗ |
| B200 | CoreWeave | $8.60 | List | HGX B200, 8-GPU node $68.80 per 8-GPU node-hour |
Pricing page ↗ |
| B200 | Together AI | $9.29 | List | HGX B200 Reserved via sales |
Pricing page ↗ |
| B200 | Crusoe | $9.65 | List | HGX B200 Listed as rising to $10.20 on Dec 1, 2026 |
Pricing page ↗ |
| B200 | Verda | $9.75 | List | B200 SXM6 Spot listed at half price |
Pricing page ↗ |
| B200 | Vast.ai | n/a | Not published | Live marketplace pricing; no fixed list | Pricing page ↗ |
| A100 | Thunder Compute | $1.09 | List Lowest listed | A100 80GB | Pricing page ↗ |
| A100 | Massed Compute | $1.38 | List | A100 SXM4 80GB, 1x A100 family from $1.35 |
Pricing page ↗ |
| A100 | Jarvis Labs | $1.49 | List | A100 80GB Spot $0.89 |
Pricing page ↗ |
| A100 | Denvr Dataworks | $2.00 | List | A100 SXM 80GB, 8x, InfiniBand | Pricing page ↗ |
| A100 | RunPod | $2.19 | List | A100 SXM 80GB, Pods A100 PCIe $2.09 |
Pricing page ↗ |
| A100 | CoreWeave | $2.70 | List | A100 80GB, 8-GPU node $21.60 per 8-GPU node-hour |
Pricing page ↗ |
| A100 | Lambda | $3.19 | List | HGX A100 80GB, 8x instance | Pricing page ↗ |
| A100 | Crusoe | n/a | Not published | Contact sales | Pricing page ↗ |
| A100 | Vast.ai | n/a | Not published | Live marketplace pricing; no fixed list | Pricing page ↗ |
| A100 | Verda | n/a | Not published | Offered; price not captured | Pricing page ↗ |
| L40S | Hydra Host | $0.70–$1.45 | Range Lowest listed | L40S, bare metal marketplace | Pricing page ↗ |
| L40S | Massed Compute | $0.97 | List | L40S 48GB, 1x | Pricing page ↗ |
| L40S | RunPod | $1.09 | List | L40S, Pods | Pricing page ↗ |
| L40S | Nebius | $1.55 | From | L40S (Intel CPU config) From price; AMD CPU config from $1.82 |
Pricing page ↗ |
| L40S | CoreWeave | $2.25 | List | L40S 48GB, 8-GPU node $18.00 per 8-GPU node-hour |
Pricing page ↗ |
| L40S | Crusoe | n/a | Not published | Contact sales | Pricing page ↗ |
| L40S | Lambda | n/a | Not published | Not listed on the on-demand page | Pricing page ↗ |
| L40S | Vast.ai | n/a | Not published | Live marketplace pricing; no fixed list | Pricing page ↗ |
Market reference
Indices and contract benchmarks (H100)
| Ornn OCPI-H100 transacted index (Nov 19, 2026) | $2.78 / GPU-hr |
| Silicon Data H100 neocloud index (Nov 20, 2026) | $3.08 / GPU-hr |
| Silicon Data H100 hyperscaler index | $7.08 / GPU-hr |
| SemiAnalysis H100 1-year contract range (Mar 2026) | $2.00–$2.70 / GPU-hr |
Transacted and contract levels are not comparable one-for-one with on-demand list prices. See On-demand vs reserved.
Want a real number?
List prices are a starting point
Quotes for multi-node, multi-month or reserved capacity are usually negotiated, and terms matter as much as the rate. Send a brief and we will bring back quotes for your exact requirement.