GPU / INDEX / H200
The same H200 costs 9.3× more on some clouds than others.
Hopper with 141GB HBM3e · incl. GH200 superchip rows. Every provider's cheapest H200 in the sample, ranked cheapest-first. On-demand $/GPU/hr. Color is temperature: cool is cheap, white-hot is expensive.
H200 · Median $/GPU/hr
$3.89/HR
Range across providers
$1.00 — $9.31/HR
§ 01 / THE TABLE
Cheapest H200 per provider.
Rows tagged ARM are GH200 superchips — an Arm-based NVIDIA Grace CPU paired with a Hopper GPU. The catalog lists no standalone Arm GPU; this is the only Arm-silicon option in the index.
| # | Provider | Variant | Region | $/GPU/hr | Spot |
|---|---|---|---|---|---|
| 01 | Mithril | H200 | US Central2 B | $1.00 | $1.00 |
| 02 | PrimeIntellect | H200 | FI | $2.00 4× | — |
| 03 | Yotta | H200 | AP East 2 | $2.10 | — |
| 04 | Lambda | GH200 ARM | Europe Central 1 | $2.29 | — |
| 05 | Vast.ai | H200 | France, FR, EU | $2.35 | — |
| 06 | Seeweb | H200 | — | $2.60 | — |
| 07 | Spheron | gh200 ARM | — | $3.02 | — |
| 08 | GMI Cloud | h200 | — | $3.20 | — |
| 09 | Massed Compute | h200-nvl | — | $3.62 | — |
| 10 | RunPod | h200-nvl | — | $3.79 | — |
| 11 | Hyperstack | h200 | — | $3.99 | — |
| 12 | Crusoe | h200 | — | $4.29 | — |
| 13 | DigitalOcean | h200 | — | $4.47 | — |
| 14 | Jarvis Labs | h200 | — | $4.59 | — |
| 15 | Nebius | H200 | EU North1 | $5.40 | $0.79 |
| 16 | OVHcloud | h200 | — | $6.20 | — |
| 17 | CoreWeave | h200 | — | $6.30 | — |
| 18 | DataCrunch | h200 | — | $6.57 | — |
| 19 | AWS | h200 | — | $7.91 | — |
| 20 | Google Cloud | h200 | — | $9.31 | — |
SOURCE: SKYPILOT CATALOG · CATALOGS/V8 · SAMPLED 2026-11-13 · PER-GPU-HR = LIST PRICE ÷ GPU COUNT
LIVE PIPELINE · 18 CLOUDS
§ 02 / MARKET SHAPE
Where the H200 market sits.
Mithril$1.00
PrimeIntellect$2.00 4×
Yotta$2.10
Lambda$2.29
Vast.ai$2.35
Seeweb$2.60
Spheron$3.02
GMI Cloud$3.20
Massed Compute$3.62
RunPod$3.79
Hyperstack$3.99
Crusoe$4.29
DigitalOcean$4.47
Jarvis Labs$4.59
Nebius$5.40
OVHcloud$6.20
CoreWeave$6.30
DataCrunch$6.57
AWS$7.91
Google Cloud$9.31
$1.00 · COOL$9.31 · WHITE-HOT
| Mithril | $1.00 per GPU/hour |
| PrimeIntellect | $2.00 per GPU/hour (from a 4-GPU node price) |
| Yotta | $2.10 per GPU/hour |
| Lambda | $2.29 per GPU/hour (Arm-based GH200 superchip) |
| Vast.ai | $2.35 per GPU/hour |
| Seeweb | $2.60 per GPU/hour |
| Spheron | $3.02 per GPU/hour (Arm-based GH200 superchip) |
| GMI Cloud | $3.20 per GPU/hour |
| Massed Compute | $3.62 per GPU/hour |
| RunPod | $3.79 per GPU/hour |
| Hyperstack | $3.99 per GPU/hour |
| Crusoe | $4.29 per GPU/hour |
| DigitalOcean | $4.47 per GPU/hour |
| Jarvis Labs | $4.59 per GPU/hour |
| Nebius | $5.40 per GPU/hour |
| OVHcloud | $6.20 per GPU/hour |
| CoreWeave | $6.30 per GPU/hour |
| DataCrunch | $6.57 per GPU/hour |
| AWS | $7.91 per GPU/hour |
| Google Cloud | $9.31 per GPU/hour |
§ / QUOTES
Running H200? Get matched quotes.
Tell us the count, duration, and region. We review every submission by hand and connect you with providers that fit — expect a reply within one business day. No spam, no obligation.
20 DIRECT PROVIDERS · MEDIAN $3.89/GPU/HR · SAMPLE 2026-11-13