Nvidia L40 cloud rental prices
4 providers list the Nvidia L40, 4 have it in stock. Prices run from $0.34 to $1.00 per GPU hour — the same card, 3.0x apart.
Prices and stock checked 2 hours ago
What it costs you to run
Estimated at the cheapest listed rate of $0.34/GPU/hr (Vast.ai), per GPU, before storage, egress or commitment discounts.
Renting at the median instead of the floor costs you an extra $208 a month per GPU. That is 46% of the bill, for identical silicon.
Every provider renting the Nvidia L40
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 |
|---|---|---|---|---|---|---|---|---|
| Vast.ai | United States of America | 1 | 2x-2x | $0.34 | — | — | In stock | GrabVast.ai — opens the provider's site |
| Lium | — | 1 | 1x-1x | $0.36 | — | — | In stock | GrabLium — opens the provider's site |
| Massed Compute | United States of America | 6 | 1x-4x | $0.88 | — | — | In stock | GrabMassed Compute — opens the provider's site |
| Hyperstack | United Kingdom | 4 | 1x-8x | $1.00 | — | — | In stock | GrabHyperstack — opens the provider's site |
Live inventory signal
The Lium marketplace publishes GPU counts, which almost nobody else does. It holds 27 Nvidia L40 GPUs at $0.36/GPU/hr, with 48% already rented. There is slack at that price today.
All 12 configurations
| Configuration | Provider | Region | VRAM | vCPU | RAM | Billing | Per GPU/hr | Total/hr | Stock |
|---|---|---|---|---|---|---|---|---|---|
| 2x L40 | Vast.ai | VN | 90 GB | 96 | 376 GB | On-Demand | $0.34 | $0.67 | In stock |
| 1x NVIDIA L40 | Lium | — | — | — | — | On-Demand | $0.36 | $0.36 | In stock |
| 1x L40 | Massed Compute | desmoines-usa-1 | 48 GB | 26 | 192 GB | On-Demand | $0.88 | $0.88 | In stock |
| 1x L40 | Massed Compute | kansascity-usa-1 | 48 GB | 26 | 192 GB | On-Demand | $0.88 | $0.88 | Out of stock |
| 2x L40 | Massed Compute | desmoines-usa-1 | 96 GB | 50 | 384 GB | On-Demand | $0.99 | $1.98 | In stock |
| 2x L40 | Massed Compute | kansascity-usa-1 | 96 GB | 50 | 384 GB | On-Demand | $0.99 | $1.98 | Out of stock |
| 8x L40 | Hyperstack | montreal-canada-2 | 384 GB | 252 | 464 GB | On-Demand | $1.00 | $8.00 | In stock |
| 1x L40 | Hyperstack | montreal-canada-2 | 48 GB | 28 | 58 GB | On-Demand | $1.00 | $1.00 | In stock |
| 2x L40 | Hyperstack | montreal-canada-2 | 96 GB | 60 | 116 GB | On-Demand | $1.00 | $2.00 | In stock |
| 4x L40 | Hyperstack | montreal-canada-2 | 192 GB | 126 | 232 GB | On-Demand | $1.00 | $4.00 | In stock |
| 4x L40 | Massed Compute | desmoines-usa-1 | 192 GB | 100 | 768 GB | On-Demand | $1.49 | $5.96 | In stock |
| 4x L40 | Massed Compute | kansascity-usa-1 | 192 GB | 100 | 768 GB | On-Demand | $1.49 | $5.96 | Out of stock |
What fits in 48 GB
The question behind most rentals: will the model load. Sized against one Nvidia L40.
| Model | 16-bit | 8-bit | 4-bit |
|---|---|---|---|
| DeepSeek-V3 DeepSeek · 671B params · 37B active | 1,610 GB 34× · 22 GB free | 805 GB 17× · 11 GB free | 403 GB 9× · 29 GB free |
| Llama 3.1 405B Meta · 405B params | 972 GB 21× · 36 GB free | 486 GB 11× · 42 GB free | 243 GB 6× · 45 GB free |
| Qwen3 235B-A22B Alibaba · 235B params · 22B active | 564 GB 12× · 12 GB free | 282 GB 6× · 6 GB free | 141 GB 3× · 3 GB free |
| Mixtral 8x22B Mistral · 141B params · 39B active | 338 GB 8× · 46 GB free | 169 GB 4× · 23 GB free | 85 GB 2× · 11 GB free |
| Mistral Large 2 Mistral · 123B params | 295 GB 7× · 41 GB free | 148 GB 4× · 44 GB free | 74 GB 2× · 22 GB free |
| gpt-oss-120b OpenAI · 117B params · 5.1B active | 281 GB 6× · 7 GB free | 140 GB 3× · 4 GB free | 70 GB 2× · 26 GB free |
| Qwen2.5 72B Alibaba · 72.7B params | 174 GB 4× · 18 GB free | 87 GB 2× · 9 GB free | 44 GB 1× · 4 GB free |
| Llama 3.3 70B Meta · 70.6B params | 169 GB 4× · 23 GB free | 85 GB 2× · 11 GB free | 42 GB 1× · 6 GB free |
| Qwen3 32B Alibaba · 32.8B params | 79 GB 2× · 17 GB free | 39 GB 1× · 9 GB free | 20 GB 1× · 28 GB free |
| Gemma 2 27B Google · 27.2B params | 65 GB 2× · 31 GB free | 33 GB 1× · 15 GB free | 16 GB 1× · 32 GB free |
| Mistral Small 3 Mistral · 24B params | 58 GB 2× · 38 GB free | 29 GB 1× · 19 GB free | 14 GB 1× · 34 GB free |
| Llama 3.1 8B Meta · 8B params | 19 GB 1× · 29 GB free | 10 GB 1× · 38 GB free | 5 GB 1× · 43 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 L40.
Nvidia L40 specifications
| Architecture | Ada Lovelace |
|---|---|
| Memory per GPU | 48 GB GDDR6 |
| Memory bandwidth | 864 GB/s |
| Release date | Q4 2022 |
| FP8 compute rate | 362 TFLOPS (dense), 724 with sparsity |
| FP16 / BF16 compute rate | 181.1 TFLOPS (dense), 362.1 with sparsity |
| INT8 compute rate | 362 TOPS (dense), 724 with sparsity |
| Process node | TSMC 4N |
| Board power | 300 W |
Frequently asked questions
How much does it cost to rent a Nvidia L40 per hour?
As of September 6, 2026 the Nvidia L40 rents from $0.34 per GPU per hour at Vast.ai. The median across 4 providers is $0.62 and the dearest listing asks $1.00.
Who is the cheapest Nvidia L40 provider right now?
Vast.ai at $0.34 per GPU per hour. That is 46% under the median of $0.62, so renting from a mid-priced provider costs you $0.28 an hour more for the same silicon.
What does a Nvidia L40 cost per month?
At the cheapest listed rate of $0.34 per GPU per hour, one Nvidia L40 running non-stop for a 730-hour month costs $245. At the median rate it costs $453.
Can I actually get a Nvidia L40 today?
Yes. 4 of 4 providers reported Nvidia L40 stock at the last check. The cheapest one with confirmed stock is Vast.ai at $0.34 per GPU per hour.
How much VRAM does the Nvidia L40 have?
48 GB GDDR6 per board. That holds a model of roughly 19 billion parameters at 16-bit, or about 38 billion at 8-bit, before context and activations.
What architecture is the Nvidia L40, and when did it launch?
Ada Lovelace, released Q4 2022. 4 cloud providers publish a price for it.