Cloud GPUs

GPU cloud for AI training, fine-tuning and inference

Run AI and other GPU-heavy work on GPU instances and Kubernetes GPU node pools, with NVMe storage, private networking and billing in rupees. Tell us what you need and we'll size it with you.

Talk to our teamHow it works
Pricing depends on the GPU, how many you need and for how long, so every quote is made for your workload.
GPU node4 GPUs · NVMe scratch volumeRunningGPU 094%GPU 189%GPU 296%GPU 383%Training lossepoch 12 / 40GPU

Workloads

What you can run

From a single GPU for inference to multi-GPU instances for training.

Training

Train models from scratch on multi-GPU instances, with fast NVMe volumes for datasets and checkpoints.

Fine-tuning

Adapt open models to your own data on one or a few GPUs, and keep that data inside your own environment.

Inference

Serve models behind a load balancer, on GPU instances or on Kubernetes GPU node pools next to the rest of your app.

Rendering and simulation

Rendering, transcoding, scientific computing and other parallel jobs that run faster on GPUs.

How it works

Tell us the job, we'll size the GPUs

GPU capacity is set up with our team, then run by you from the console.

  1. Tell us what you need

    The workload, the GPU you have in mind (for example H100, A100, L40S or L4), how many and for how long.

  2. We size it with you

    An engineer checks the fit, confirms availability and sends you a price in rupees.

  3. Run it from your console

    Your GPU instances sit in the same console as the rest of your cloud. Start, stop, snapshot and watch them yourself.

Self-service

GPUs in the same console as everything else

GPU instances sit next to your servers, volumes and clusters. Watch every GPU's load and your jobs' progress yourself, and call on our team when you need it.

app.neoncloud.com/gpusGPU instances3 instances · 13 GPUstrain-llm-018 × GPU · TrainingRunningfinetune-024 × GPU · Fine-tuningRunninginfer-api-031 × GPU · InferenceRunningGPU load · train-llm-0101234567

Included

Everything around the GPUs

The same building blocks as the rest of Neon Cloud, so a GPU workload fits into your environment.

GPU virtual machines

Dedicated GPU instances you control: install the drivers, frameworks and tools your team already uses.

Kubernetes GPU node pools

Add GPU node pools to a Neon Kubernetes Engine cluster, alongside your CPU pools, and schedule GPU workloads like any other.

NVMe storage

NVMe block storage for datasets, checkpoints and model weights, and object storage for everything else.

Private networking

VPCs, security groups and load balancers, the same as the rest of your environment.

One bill in rupees

GPUs are billed with the rest of your Neon Cloud account, in INR and before GST.

Engineers who help

Help from our team while you choose a setup, get it running and once it's in production.

Cloud GPUs FAQ

Questions about Cloud GPUs

Pricing, access and which GPUs you can ask for.

Ask about GPUs

  • GPU pricing depends on the GPU, how many you need and for how long, so we quote each request rather than publish a price list. Tell us about your workload through the contact form or at sales@neoncloud.com, and we'll reply with availability and a price in rupees.

Tell us what you want to run

Share your workload, the GPU you have in mind and how many, and our team will come back with availability and a price.

Talk to our teamSee all prices