GPU model · price + stock · hourly snapshot
Nvidia GH200 cloud rental prices
2 providers list the Nvidia GH200, 1 has it in stock. Prices run from $2.29 to $4.23 per GPU hour — the same card, 1.8x apart.
LIVE Prices and stock checked 1 minute ago
Providers
Every provider renting the Nvidia GH200
| Provider | From | Stock | Grab |
|---|---|---|---|
| Lambda Labs | $2.29 | In stock | Grab |
| Latitude.sh | $4.23 | Out of stock | Grab |
In stock first, then cheapest. Stock is what the provider reported at our last check. One GPU for a month (730h) at $2.29: $1,672.
The Nvidia GH200 sells out fast. Get one email when it is back in stock.
Price history
Price trend
$2.29 /GPU/hr · Oct 22
One point per day: the lowest listed $/GPU/hr across every provider we track, captured daily. 4 offers in the latest snapshot.
Configurations
All 2 configurations
| Configuration | VRAM | Per GPU/hr | Stock |
|---|---|---|---|
| 1x GH200Lambda Labs | 96 GB | $2.29 | In stock |
| 1x GH200Latitude.sh | 96 GB | $4.23 | Out of stock |
Model fit
What fits in 96 GB
The question behind most rentals: will the model load. Sized against one Nvidia GH200.
| Model | 16-bit | 8-bit | 4-bit |
|---|---|---|---|
| DeepSeek-V3 DeepSeek · 671B params · 37B active | 1,610 GB 17× · 22 GB free | 805 GB 9× · 59 GB free | 403 GB 5× · 77 GB free |
| Llama 3.1 405B Meta · 405B params | 972 GB 11× · 84 GB free | 486 GB 6× · 90 GB free | 243 GB 3× · 45 GB free |
| Qwen3 235B-A22B Alibaba · 235B params · 22B active | 564 GB 6× · 12 GB free | 282 GB 3× · 6 GB free | 141 GB 2× · 51 GB free |
| Mixtral 8x22B Mistral · 141B params · 39B active | 338 GB 4× · 46 GB free | 169 GB 2× · 23 GB free | 85 GB 1× · 11 GB free |
| Mistral Large 2 Mistral · 123B params | 295 GB 4× · 89 GB free | 148 GB 2× · 44 GB free | 74 GB 1× · 22 GB free |
| gpt-oss-120b OpenAI · 117B params · 5.1B active | 281 GB 3× · 7 GB free | 140 GB 2× · 52 GB free | 70 GB 1× · 26 GB free |
| Qwen2.5 72B Alibaba · 72.7B params | 174 GB 2× · 18 GB free | 87 GB 1× · 9 GB free | 44 GB 1× · 52 GB free |
| Llama 3.3 70B Meta · 70.6B params | 169 GB 2× · 23 GB free | 85 GB 1× · 11 GB free | 42 GB 1× · 54 GB free |
| Qwen3 32B Alibaba · 32.8B params | 79 GB 1× · 17 GB free | 39 GB 1× · 57 GB free | 20 GB 1× · 76 GB free |
| Gemma 2 27B Google · 27.2B params | 65 GB 1× · 31 GB free | 33 GB 1× · 63 GB free | 16 GB 1× · 80 GB free |
| Mistral Small 3 Mistral · 24B params | 58 GB 1× · 38 GB free | 29 GB 1× · 67 GB free | 14 GB 1× · 82 GB free |
| Llama 3.1 8B Meta · 8B params | 19 GB 1× · 77 GB free | 10 GB 1× · 86 GB free | 5 GB 1× · 91 GB free |
Required memory is weights plus 20% for KV cache and runtime at short context; long contexts and large batches need more. Cells tinted green fit on a single Nvidia GH200.
Specifications
Nvidia GH200 specifications
| Architecture | Grace Hopper |
|---|---|
| Memory per GPU | 96 GB HBM3 |
| Memory bandwidth | 4,000 GB/s |
| Release date | Q2 2023 |
| GPU interconnect | NVLink 900 GB/s |
| FP8 compute rate | 1,979 TFLOPS (dense), 3,958 with sparsity |
| FP16 / BF16 compute rate | 989.5 TFLOPS (dense), 1,979 with sparsity |
| INT8 compute rate | 1,979 TOPS (dense), 3,958 with sparsity |
| Process node | TSMC 4N |
Alternatives