Versatile performance
Handle diverse workloads on a single architecture, from LLM inference and generative AI to 3D rendering and video encoding.
L4 GPU Instance
Affordable inference.
Cost-effective entry GPU
Start building your AI infrastructure from just €0.79/GPU/hour, making the L4 the ideal entry point for start-ups prototyping without upfront costs.
Modern architecture
Take advantage of faster inference and rendering with Ada Lovelace’s architecture and 24GB GDDR6 memory.
The multi-purpose GPU for every workload
Whether you’re serving AI models, processing video streams, or rendering 3D scenes, the L4 GPU Instance provides a balanced, budget-friendly foundation. It eliminates the need to pay unnecessary GPU capacity for your less demanding tasks, giving you the right performance.
Deploy the latest size-optimized Large Language Models (LLMs) and Diffusion models at scale without the massive power overhead of flagship training GPUs.
Process video at scale with dedicated hardware encoders. With native support for AV1 encoding and up to 3x the performance of T4 for H264, the L4 is optimized for transcoding, live streaming, and real-time computer vision applications.
Accelerate industrial digital twins and cloud gaming. Third-generation RT Cores provide real-time ray tracing capabilities, enabling engineering teams to visualize complex 3D models and simulations faster.
Specifications
View pricing- gpu
GPU
NVIDIA L4 Tensor Core.
- processor_type
Architecture
NVIDIA Lovelace 2022.
- bandwidth
VRAM
24 GB GDDR6 per GPU (300GB/s).
- processor
CPU
8-64 vCPUs AMD EPYC™ 7413.
- processor_frequency
Processor frequency
2.65 Ghz.
- memory
RAM
48-384 GB.
- memory_type
RAM type
DDR4.
- bandwidth
Network bandwidth
Up to 20 Gbps.
- storage
Storage
Block Storage.
- threads_cores
GPU performance
Tensor Cores 4th generation, RT Cores 3rd generation.
- service_level
SLA
99.5%.
Estimate your GPU costs
Choose your plan
Estimated cost
| Option and value | Price |
|---|---|
| ZoneParis 2 | |
| Instance1x | 0€ |
| Volume10GB | 0€ |
| Flexible IPv4No | 0€ |
100% renewable energy, up to 30% less power
DC5 (PAR2) is one of Europe's greenest data centers, powered entirely by renewable wind and hydro energy (GO-certified) and cooled with ultra-efficient free and adiabatic cooling. With a PUE of 1.16 (vs. the 1.55 industry average), it slashes energy use by 30% compared to traditional data centers.
Looking for more power? Discover our full range.
Managed Inference
Deploy AI models in a dedicated inference infrastructure, with tailored security and predictable throughput.
L40S GPUs
Accelerate the next generation of AI-enabled applications with the universal L40S GPU Instance.
H100 PCIe GPUs
Accelerate AI applications with H100 GPUs.
Choose the cloud built for what's next
Customer data sovereignty
Dependency is the enemy of resilience. Customers want their data hosted by a regional provider. Gain sovereignty with our multi-cloud tools & infrastructure.
Sustainable data centers
We recycle our hardware, only use renewable energy and pay close attention to our water usage. Also, our Power Usage Effectiveness (PUE) is displayed online 24/7 for you to see for yourself.
Low latency
Every complete cloud ecosystem needs 100% reliability, which is why we provide nine Availability Zones in three different regions.
Frequently asked questions
What is included in the Instance price?
Our GPU Instances' prices include the vCPU and RAM needed for optimal performance.
To launch the L4 GPU Instance you will need to provision a minimum of Block Storage and a flexible IP at your expense.
Any doubt about the price, use the calculator, it's made for it!
What to consider before choosing a cloud GPU?
Before you rent GPU infrastructure, you need to map your technical requirements to the right instance type to ensure you aren't over-provisioning. Keep the following factors in mind:
Primary use case: are you running real-time video encoding, entry-level AI inference, or basic graphics processing where extreme compute power is unnecessary?
Cost vs. performance: evaluate whether your application actually requires flagship hardware, or if a cost-optimized architecture will deliver the same end-user latency.
Memory constraints: check the size of the models you intend to serve. For smaller models or video streams, 24GB of VRAM is often suitable.
CPU and RAM balance: make sure the attached system resources can handle your web traffic or data preprocessing without bottlenecking your cloud GPU.
To help you make the right technical decision, read our comprehensive guide on evaluating GPU instances here