Built for video and light inference
With 2 NVENC and 4 NVDEC engines and a 72 W power draw, the L4 suits transcoding, speech-to-text and embedding services that run all day.
GPU rental prices · Ada Lovelace
NVIDIA L4 rental prices
NVIDIA L4 rental starts at $0.59 per GPU-hour (Runpod, PCIe, 24 GB) across the 7 offers we track from 3 providers. The typical price is $1.00, and one GPU running all month costs about $431 at the lowest rate. A 72 W, low-profile card for small models, speech, embeddings and video. Its 300 GB/s of memory bandwidth limits how fast it generates text, so it suits small models best.
- Lowest per GPU-hour
- $0.59 Runpod
- Typical per GPU-hour
- $1.00 median, USD
- One GPU all month
- $431 at the lowest rate
- Providers
- 3 7 offers
On-demand prices
NVIDIA L4 price per GPU-hour
Every L4 offer in our catalog, cheapest first. Multi-GPU machines are divided by their GPU count, so you can compare one GPU with one GPU. Monthly cost assumes the GPU runs all 730 hours.
| Provider and offer | Version | GPU memory | GPUs | Per GPU-hour | Per month, 1 GPU | Checked |
|---|---|---|---|---|---|---|
| RunpodRunpod Pod · NVIDIA L4 | PCIe, 24 GB | 24 GB | 1 | $0.59 | $431 | 3 Nov 2026 |
| OVHcloudOVHcloud US l4-180 | PCIe, 24 GB | 24 GB | 2$2.000/hr total | $1.00 | $730 | 3 Nov 2026 |
| OVHcloudOVHcloud US l4-90 | PCIe, 24 GB | 24 GB | 1 | $1.00 | $730 | 3 Nov 2026 |
| ScalewayScaleway L4-8-24G | PCIe, 24 GB | 24 GB | 8€6.300/hr total | €0.788 | €575 | 3 Nov 2026 |
| ScalewayScaleway L4-4-24G | PCIe, 24 GB | 24 GB | 4€3.150/hr total | €0.788 | €575 | 3 Nov 2026 |
| ScalewayScaleway L4-1-24G | PCIe, 24 GB | 24 GB | 1 | €0.79 | €577 | 3 Nov 2026 |
| ScalewayScaleway L4-2-24G | PCIe, 24 GB | 24 GB | 2€1.580/hr total | €0.79 | €577 | 3 Nov 2026 |
On-demand list prices before tax, from each provider's pricing page. Prices in euros are not converted and are left out of the lowest and typical figures. Community and reserved prices, where a provider offers them, are lower; see the notes below. Full offer details, storage and billing terms.
Before you rent
What changes the price
Euro prices
Scaleway bills in euros (€0.79 per GPU-hour), so its offers are listed after the USD ones and left out of the lowest and typical figures.
Community prices
Runpod's Community Cloud lists the L4 at $0.44 an hour, against $0.59 on its Secure Cloud.
AI models
Which models fit on the L4
Memory to load each model size on 24 GB GPUs, with a 20% allowance for runtime buffers, and what that many GPUs cost per hour at the lowest rate above. Long contexts and many parallel users need more memory. More than one GPU only works if your inference server splits the model across them.
| Model size | 4-bit | 8-bit | 16-bit |
|---|---|---|---|
| 8BLlama 3.1 8B, Qwen3 8B | 4.5 GB1 GPU · $0.59/hr | 9.0 GB1 GPU · $0.59/hr | 17.9 GB1 GPU · $0.59/hr |
| 32BQwen3 32B, Gemma 4 31B | 17.9 GB1 GPU · $0.59/hr | 35.8 GB2 GPUs · $1.18/hr | 71.6 GB3 GPUs · $1.77/hr |
| 70BLlama 3.3 70B, Qwen 72B | 39.2 GB2 GPUs · $1.18/hr | 78.3 GB4 GPUs · $2.36/hr | 156.5 GB7 GPUs · $4.13/hr |
| 120Bgpt-oss-120b class | 67.1 GB3 GPUs · $1.77/hr | 134.2 GB6 GPUs · $3.54/hr | 268.3 GB12 GPUs · more than one 8-GPU server |
| 685BDeepSeek-V3.2 | 382.8 GB16 GPUs · more than one 8-GPU server | 765.6 GB32 GPUs · more than one 8-GPU server | 1,531.1 GB64 GPUs · more than one 8-GPU server |
Weights only, from the parameter count. Check a specific model, context length and user count in the model hardware calculator.
Hardware
NVIDIA L4 specifications
| Specification | L4 |
|---|---|
| GPU memory | 24 GB |
| Memory bandwidth | 300 GB/s |
| FP8 Tensor (sparse) | 485 TFLOPS |
| Max power | 72 W |
| Form factor | PCIe, single-slot, low-profile |
| Video engines | 2 NVENC, 4 NVDEC, 4 JPEG decoders |
From NVIDIA's product pages. Tensor figures marked “sparse” assume structured sparsity; dense throughput is half.
Questions
L4 rental FAQ
How much does it cost to rent an L4?
$0.59 per GPU-hour is the lowest on-demand price we track, from Runpod. The typical price is $1.00, and one GPU running all month costs about $431 at the lowest rate.
Can an L4 run Llama 3.1 8B?
Yes. An 8B model needs about 9.0 GB at 8-bit and 17.9 GB at 16-bit with a 20% allowance, both within 24 GB.
Should I rent an L4 or an L40S?
The L40S has twice the memory (48 GB against 24 GB) and almost three times the bandwidth (864 GB/s against 300 GB/s), so it runs larger models and generates text faster. The L4 costs less per hour and draws 72 W, which makes it the cheaper choice for small models and video work.
Compare
Other GPU rental prices
Prices and availability change, and we don't check stock. Confirm the current price and region with the provider before you rent. Sources: www.nvidia.com, www.runpod.io, us.ovhcloud.com, www.scaleway.com . See our affiliate disclosure.