Nvidia B200 cloud rental prices
7 providers list the Nvidia B200, 3 have it in stock. Prices run from $3.74 to $6.69 per GPU hour — the same card, 1.8x apart.
Prices and stock checked 0 minutes ago
What it costs you to run
Estimated at the cheapest listed rate of $4.25/GPU/hr (Lium), per GPU, before storage, egress or commitment discounts.
Renting at the median instead of the floor costs you an extra $1,149 a month per GPU. That is 30% of the bill, for identical silicon.
Every provider renting the Nvidia B200
Ordered strictly by entry price per GPU hour. Rows above the median are marked. Stock is what the provider reported at our last check, not a guarantee.
| Provider | Country | Configs | Scale | From | Spot | Reserved | Stock | Grab |
|---|---|---|---|---|---|---|---|---|
| BoostRun | — | 2 | 8x-8x | $3.74 | — | — | Out of stock | GrabBoostRun — opens the provider's site |
| Lium | — | 1 | 1x-1x | $4.25 | — | — | In stock | GrabLium — opens the provider's site |
| Excess Supply | — | 1 | 8x-8x | $4.51 | — | — | Out of stock | GrabExcess Supply — opens the provider's site |
| Vast.ai | United States of America | 1 | 2x-2x | $5.31 | — | — | In stock | GrabVast.ai — opens the provider's site |
| Runpod | United States of America | 2 | 1x-1x | $5.98 | $5.98 | — | In stock | GrabRunpod — opens the provider's site |
| Verda | Finland | 4 | 1x-8x | $6.30 | — | — | Out of stock | GrabVerda — opens the provider's site |
| Lambda Labs | United States of America | 8 | 1x-8x | $6.69 | — | — | Out of stock | GrabLambda Labs — opens the provider's site |
Live inventory signal
The Lium marketplace publishes GPU counts, which almost nobody else does. It holds 57 Nvidia B200 GPUs at $4.25/GPU/hr, with 67% already rented. There is slack at that price today.
All 19 configurations
| Configuration | Provider | Region | VRAM | vCPU | RAM | Billing | Per GPU/hr | Total/hr | Stock |
|---|---|---|---|---|---|---|---|---|---|
| 8x B200 | BoostRun | dallas-usa-4 | 1536 GB | 96 | 2048 GB | On-Demand | $3.74 | $29.92 | Out of stock |
| 8x B200 | BoostRun | charlotte-usa-1 | 1536 GB | 96 | 2048 GB | On-Demand | $3.74 | $29.92 | Out of stock |
| 1x NVIDIA B200 | Lium | — | — | — | — | On-Demand | $4.25 | $4.25 | In stock |
| 8x B200 | Excess Supply | kansascity-usa-4 | 1536 GB | 160 | 1792 GB | On-Demand | $4.51 | $36.11 | Out of stock |
| 2x B200 | Vast.ai | US | 358 GB | 48 | 499 GB | On-Demand | $5.31 | $10.63 | In stock |
| 1x B200 | Runpod | — | 180 GB | — | — | On-Demand | $5.98 | $5.98 | In stock |
| 1x B200 | Runpod | — | 180 GB | — | — | Spot | $5.98 | $5.98 | In stock |
| 8x B200 | Verda | helsinki-finland-5 | 1536 GB | 248 | 1360 GB | On-Demand | $6.30 | $50.44 | Out of stock |
| 4x B200 | Verda | helsinki-finland-5 | 768 GB | 124 | 680 GB | On-Demand | $6.42 | $25.67 | Out of stock |
| 1x B200 | Verda | helsinki-finland-5 | 192 GB | 31 | 170 GB | On-Demand | $6.52 | $6.52 | Out of stock |
| 2x B200 | Verda | helsinki-finland-5 | 384 GB | 62 | 340 GB | On-Demand | $6.52 | $13.04 | Out of stock |
| 8x B200 | Lambda Labs | sydney-australia-2 | 1536 GB | 208 | 2900 GB | On-Demand | $6.69 | $53.52 | Out of stock |
| 8x B200 | Lambda Labs | dulles-usa-1 | 1536 GB | 208 | 2900 GB | On-Demand | $6.69 | $53.52 | Out of stock |
| 4x B200 | Lambda Labs | sydney-australia-2 | 768 GB | 104 | 1440 GB | On-Demand | $6.79 | $27.16 | Out of stock |
| 4x B200 | Lambda Labs | dulles-usa-1 | 768 GB | 104 | 1440 GB | On-Demand | $6.79 | $27.16 | Out of stock |
| 2x B200 | Lambda Labs | sydney-australia-2 | 384 GB | 52 | 720 GB | On-Demand | $6.89 | $13.78 | Out of stock |
| 2x B200 | Lambda Labs | dulles-usa-1 | 384 GB | 52 | 720 GB | On-Demand | $6.89 | $13.78 | Out of stock |
| 1x B200 | Lambda Labs | dulles-usa-1 | 192 GB | 26 | 360 GB | On-Demand | $6.99 | $6.99 | Out of stock |
| 1x B200 | Lambda Labs | sydney-australia-2 | 192 GB | 26 | 360 GB | On-Demand | $6.99 | $6.99 | Out of stock |
What fits in 180 GB
The question behind most rentals: will the model load. Sized against one Nvidia B200.
| Model | 16-bit | 8-bit | 4-bit |
|---|---|---|---|
| DeepSeek-V3 DeepSeek · 671B params · 37B active | 1,610 GB 9× · 10 GB free | 805 GB 5× · 95 GB free | 403 GB 3× · 137 GB free |
| Llama 3.1 405B Meta · 405B params | 972 GB 6× · 108 GB free | 486 GB 3× · 54 GB free | 243 GB 2× · 117 GB free |
| Qwen3 235B-A22B Alibaba · 235B params · 22B active | 564 GB 4× · 156 GB free | 282 GB 2× · 78 GB free | 141 GB 1× · 39 GB free |
| Mixtral 8x22B Mistral · 141B params · 39B active | 338 GB 2× · 22 GB free | 169 GB 1× · 11 GB free | 85 GB 1× · 95 GB free |
| Mistral Large 2 Mistral · 123B params | 295 GB 2× · 65 GB free | 148 GB 1× · 32 GB free | 74 GB 1× · 106 GB free |
| gpt-oss-120b OpenAI · 117B params · 5.1B active | 281 GB 2× · 79 GB free | 140 GB 1× · 40 GB free | 70 GB 1× · 110 GB free |
| Qwen2.5 72B Alibaba · 72.7B params | 174 GB 1× · 6 GB free | 87 GB 1× · 93 GB free | 44 GB 1× · 136 GB free |
| Llama 3.3 70B Meta · 70.6B params | 169 GB 1× · 11 GB free | 85 GB 1× · 95 GB free | 42 GB 1× · 138 GB free |
| Qwen3 32B Alibaba · 32.8B params | 79 GB 1× · 101 GB free | 39 GB 1× · 141 GB free | 20 GB 1× · 160 GB free |
| Gemma 2 27B Google · 27.2B params | 65 GB 1× · 115 GB free | 33 GB 1× · 147 GB free | 16 GB 1× · 164 GB free |
| Mistral Small 3 Mistral · 24B params | 58 GB 1× · 122 GB free | 29 GB 1× · 151 GB free | 14 GB 1× · 166 GB free |
| Llama 3.1 8B Meta · 8B params | 19 GB 1× · 161 GB free | 10 GB 1× · 170 GB free | 5 GB 1× · 175 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 B200.
Nvidia B200 specifications
| Architecture | Blackwell |
|---|---|
| Memory per GPU | 180 GB HBM3e |
| Memory bandwidth | 7,700 GB/s |
| Release date | Q1 2024 |
| GPU interconnect | NVLink 1,800 GB/s |
| FP4 compute rate | 9,000 TFLOPS (dense), 18,000 with sparsity |
| FP8 compute rate | 4,500 TFLOPS (dense), 9,000 with sparsity |
| FP16 / BF16 compute rate | 2,250 TFLOPS (dense), 4,500 with sparsity |
| INT8 compute rate | 4,500 TOPS (dense), 9,000 with sparsity |
| Process node | TSMC 4NP |
| Board power | 1,000 W |
| Partitioning | MIG, up to 7 instances |
Frequently asked questions
How much does it cost to rent a Nvidia B200 per hour?
As of September 6, 2026 the Nvidia B200 rents from $3.74 per GPU per hour at BoostRun. The median across 7 providers is $5.31 and the dearest listing asks $6.69.
Who is the cheapest Nvidia B200 provider right now?
BoostRun at $3.74 per GPU per hour. That is 30% under the median of $5.31, so renting from a mid-priced provider costs you $1.57 an hour more for the same silicon.
What does a Nvidia B200 cost per month?
At the cheapest listed rate of $3.74 per GPU per hour, one Nvidia B200 running non-stop for a 730-hour month costs $2,730. At the median rate it costs $3,879.
Can I actually get a Nvidia B200 today?
Yes. 3 of 7 providers reported Nvidia B200 stock at the last check. The cheapest one with confirmed stock is Lium at $4.25 per GPU per hour.
How much VRAM does the Nvidia B200 have?
180 GB HBM3e per board. That holds a model of roughly 72 billion parameters at 16-bit, or about 144 billion at 8-bit, before context and activations.
What architecture is the Nvidia B200, and when did it launch?
Blackwell, released Q1 2024. 7 cloud providers publish a price for it.