Nvidia L4 cloud rental prices
4 providers list the Nvidia L4, 3 have it in stock. Prices run from $0.11 to $0.95 per GPU hour — the same card, 8.6x apart.
Prices and stock checked 2 hours ago
What it costs you to run
Estimated at the cheapest listed rate of $0.13/GPU/hr (Vast.ai), per GPU, before storage, egress or commitment discounts.
Renting at the median instead of the floor costs you an extra $129 a month per GPU. That is 62% of the bill, for identical silicon.
Every provider renting the Nvidia L4
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 |
|---|---|---|---|---|---|---|---|---|
| Lium | — | 1 | 1x-1x | $0.11 | — | — | Out of stock | GrabLium — opens the provider's site |
| Vast.ai | United States of America | 1 | 1x-1x | $0.13 | — | — | In stock | GrabVast.ai — opens the provider's site |
| Runpod | United States of America | 2 | 1x-1x | $0.44 | $0.44 | — | In stock | GrabRunpod — opens the provider's site |
| Scaleway | France | 12 | 1x-8x | $0.95 | — | — | In stock | GrabScaleway — opens the provider's site |
Live inventory signal
The Lium marketplace publishes GPU counts, which almost nobody else does. It holds 0 Nvidia L4 GPUs at $0.11/GPU/hr, with 0% already rented. There is slack at that price today.
All 16 configurations
| Configuration | Provider | Region | VRAM | vCPU | RAM | Billing | Per GPU/hr | Total/hr | Stock |
|---|---|---|---|---|---|---|---|---|---|
| 1x NVIDIA L4 | Lium | — | — | — | — | On-Demand | $0.11 | $0.11 | Out of stock |
| 1x L4 | Vast.ai | BE | 22 GB | 8 | 31 GB | On-Demand | $0.13 | $0.13 | In stock |
| 1x L4 | Runpod | — | 24 GB | — | — | On-Demand | $0.44 | $0.44 | In stock |
| 1x L4 | Runpod | — | 24 GB | — | — | Spot | $0.44 | $0.44 | In stock |
| 1x L4 | Scaleway | paris-france-5 | 24 GB | 8 | 48 GB | On-Demand | $0.95 | $0.95 | Out of stock |
| 1x L4 | Scaleway | paris-france-1 | 24 GB | 8 | 48 GB | On-Demand | $0.95 | $0.95 | Out of stock |
| 1x L4 | Scaleway | warsaw-poland-1 | 24 GB | 8 | 48 GB | On-Demand | $0.95 | $0.95 | In stock |
| 4x L4 | Scaleway | paris-france-5 | 96 GB | 32 | 192 GB | On-Demand | $0.95 | $3.80 | Out of stock |
| 4x L4 | Scaleway | warsaw-poland-1 | 96 GB | 32 | 192 GB | On-Demand | $0.95 | $3.80 | In stock |
| 4x L4 | Scaleway | paris-france-1 | 96 GB | 32 | 192 GB | On-Demand | $0.95 | $3.80 | Out of stock |
| 2x L4 | Scaleway | paris-france-5 | 48 GB | 16 | 96 GB | On-Demand | $0.95 | $1.90 | Out of stock |
| 2x L4 | Scaleway | paris-france-1 | 48 GB | 16 | 96 GB | On-Demand | $0.95 | $1.90 | Out of stock |
| 2x L4 | Scaleway | warsaw-poland-1 | 48 GB | 16 | 96 GB | On-Demand | $0.95 | $1.90 | In stock |
| 8x L4 | Scaleway | paris-france-5 | 192 GB | 64 | 384 GB | On-Demand | $0.95 | $7.60 | Out of stock |
| 8x L4 | Scaleway | paris-france-1 | 192 GB | 64 | 384 GB | On-Demand | $0.95 | $7.60 | Out of stock |
| 8x L4 | Scaleway | warsaw-poland-1 | 192 GB | 64 | 384 GB | On-Demand | $0.95 | $7.60 | In stock |
What fits in 24 GB
The question behind most rentals: will the model load. Sized against one Nvidia L4.
| Model | 16-bit | 8-bit | 4-bit |
|---|---|---|---|
| DeepSeek-V3 DeepSeek · 671B params · 37B active | 1,610 GB 68× · 22 GB free | 805 GB 34× · 11 GB free | 403 GB 17× · 5 GB free |
| Llama 3.1 405B Meta · 405B params | 972 GB 41× · 12 GB free | 486 GB 21× · 18 GB free | 243 GB 11× · 21 GB free |
| Qwen3 235B-A22B Alibaba · 235B params · 22B active | 564 GB 24× · 12 GB free | 282 GB 12× · 6 GB free | 141 GB 6× · 3 GB free |
| Mixtral 8x22B Mistral · 141B params · 39B active | 338 GB 15× · 22 GB free | 169 GB 8× · 23 GB free | 85 GB 4× · 11 GB free |
| Mistral Large 2 Mistral · 123B params | 295 GB 13× · 17 GB free | 148 GB 7× · 20 GB free | 74 GB 4× · 22 GB free |
| gpt-oss-120b OpenAI · 117B params · 5.1B active | 281 GB 12× · 7 GB free | 140 GB 6× · 4 GB free | 70 GB 3× · 2 GB free |
| Qwen2.5 72B Alibaba · 72.7B params | 174 GB 8× · 18 GB free | 87 GB 4× · 9 GB free | 44 GB 2× · 4 GB free |
| Llama 3.3 70B Meta · 70.6B params | 169 GB 8× · 23 GB free | 85 GB 4× · 11 GB free | 42 GB 2× · 6 GB free |
| Qwen3 32B Alibaba · 32.8B params | 79 GB 4× · 17 GB free | 39 GB 2× · 9 GB free | 20 GB 1× · 4 GB free |
| Gemma 2 27B Google · 27.2B params | 65 GB 3× · 7 GB free | 33 GB 2× · 15 GB free | 16 GB 1× · 8 GB free |
| Mistral Small 3 Mistral · 24B params | 58 GB 3× · 14 GB free | 29 GB 2× · 19 GB free | 14 GB 1× · 10 GB free |
| Llama 3.1 8B Meta · 8B params | 19 GB 1× · 5 GB free | 10 GB 1× · 14 GB free | 5 GB 1× · 19 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 L4.
Nvidia L4 specifications
| Architecture | Ada Lovelace |
|---|---|
| Memory per GPU | 24 GB GDDR6 |
| Memory bandwidth | 300 GB/s |
| Release date | Q1 2023 |
| FP8 compute rate | 242 TFLOPS (dense), 485 with sparsity |
| FP16 / BF16 compute rate | 121 TFLOPS (dense), 242 with sparsity |
| INT8 compute rate | 242 TOPS (dense), 485 with sparsity |
| Process node | TSMC 4N |
| Board power | 72 W |
Frequently asked questions
How much does it cost to rent a Nvidia L4 per hour?
As of September 6, 2026 the Nvidia L4 rents from $0.11 per GPU per hour at Lium. The median across 4 providers is $0.29 and the dearest listing asks $0.95.
Who is the cheapest Nvidia L4 provider right now?
Lium at $0.11 per GPU per hour. That is 62% under the median of $0.29, so renting from a mid-priced provider costs you $0.18 an hour more for the same silicon.
What does a Nvidia L4 cost per month?
At the cheapest listed rate of $0.11 per GPU per hour, one Nvidia L4 running non-stop for a 730-hour month costs $80. At the median rate it costs $210.
Can I actually get a Nvidia L4 today?
Yes. 3 of 4 providers reported Nvidia L4 stock at the last check. The cheapest one with confirmed stock is Vast.ai at $0.13 per GPU per hour.
How much VRAM does the Nvidia L4 have?
24 GB GDDR6 per board. That holds a model of roughly 9 billion parameters at 16-bit, or about 19 billion at 8-bit, before context and activations.
What architecture is the Nvidia L4, and when did it launch?
Ada Lovelace, released Q1 2023. 4 cloud providers publish a price for it.