NVIDIA L4, RTX PRO 6000 and H200 GPUs in our Mumbai data centre, from ₹37,499/month. For AI inference, fine-tuning and rendering, with your data in India, a GST invoice, and engineers who check your model fits before you pay.
- ₹37,499
- from, per month
- 3
- GPUs: L4, RTX PRO 6000, H200
- 141 GB
- GPU memory on the H200
- Mumbai
- your data stays in India
- Since 2006, Jaipur
- 3,500+ businesses served
- Mumbai data centre, Yotta NM1
- GST invoice in INR
Who a GPU server is for, and who should look elsewhere
This is for you if
- You run AI in production — a chatbot, document extraction, speech-to-text, search — and need a GPU that's on all month.
- You fine-tune open models, or render 3D and video, and want the hardware in India with a GST invoice.
- Your clients want their data kept in India, and you'd rather call an engineer than open a ticket with a hyperscaler.
This is not for you if
- You need a GPU for a few hours a week → a monthly server will cost more than you need; ask about hourly billing (on request)
- You're training a large model from scratch across many GPUs → that needs a cluster with fast links between GPUs; tell us the job and we'll say plainly whether we can help
- Your work doesn't use a GPU — websites, databases, an ERP → dedicated servers or VPS
GPU server plans and prices
| Plan | GPU | vCPU | RAM | Best for | Price | Enquire |
|---|---|---|---|---|---|---|
L4 ×1 · 8 vCPU | 1× NVIDIA L4 24 GB | 8 | 32 GB | Inference, speech and video | ₹37,499 a month, ex-GST | Ask |
L4 ×1 · 16 vCPU | 1× NVIDIA L4 24 GB | 16 | 64 GB | Inference, speech and video | ₹40,699 a month, ex-GST | Ask |
L4 ×2 · 16 vCPU | 2× NVIDIA L4 48 GB | 16 | 64 GB | Inference, speech and video | ₹69,999 a month, ex-GST | Ask |
L4 ×2 · 32 vCPU | 2× NVIDIA L4 48 GB | 32 | 128 GB | Inference, speech and video | ₹80,099 a month, ex-GST | Ask |
L4 ×4 · 32 vCPU | 4× NVIDIA L4 96 GB | 32 | 128 GB | Inference, speech and video | ₹1,38,499 a month, ex-GST | Ask |
L4 ×4 · 64 vCPU | 4× NVIDIA L4 96 GB | 64 | 256 GB | Inference, speech and video | ₹1,59,299 a month, ex-GST | Ask |
RTX PRO 6000 | 1× NVIDIA RTX PRO 6000 96 GB | 64 | 192 GB | Rendering, 3D, mid-size models | ₹1,35,099 a month, ex-GST | Ask |
H200 | 1× NVIDIA H200 141 GB | 64 | 256 GB | Large models, fine-tuning | ₹2,79,399 a month, ex-GST | Ask |
Prices a month in INR, exclusive of 18% GST. Every plan includes 100 GB of NVMe storage, 10 TB of transfer a month and one IPv4 address. Hourly billing: on request. Renews at the same price. Detailed pages: L4 · RTX PRO 6000 · H200
How much does GPU cloud cost in India?
At Chandnas, a GPU server in Mumbai costs ₹37,499 a month with one NVIDIA L4, ₹1,35,099 with an RTX PRO 6000 and ₹2,79,399 with an H200, exclusive of GST. Running all month, that's about ₹51, ₹185 and ₹383 an hour.
All prices on one pageWhat moves the price
- Which GPUGPU memory decides what fits: 24 GB on an L4, 96 GB on an RTX PRO 6000, 141 GB on an H200. Buy the memory your model needs, not the fastest chip.
- How many GPUs, and the server around themL4 plans come with one, two or four GPUs, and more vCPU and RAM for loading and preparing data.
- Monthly or hourlyMonthly is cheapest for a GPU that's busy every day. For occasional jobs, ask about hourly billing.
What we do, what you do
| Task | We do | You do | Optional |
|---|---|---|---|
| GPU hardware, network and power; replacing failed parts | |||
| Installing the operating system | |||
| GPU drivers, CUDA and PyTorch | on request | keeping them updated | |
| Security updates and firewall | |||
| Monitoring GPU use and health | managed operations (quoted) | ||
| Backups of models and data | cloud backup | ||
| Your models, code and data |
- Before you pay, tell us the model and the traffic: we'll check it fits the GPU memory you're buying.
- First response target: 10 minutes if a server is down, from an engineer.
- GPU and hardware faults handled 24/7.
- If you'd rather not run the stack yourself, we can take on drivers, updates and monitoring — quoted to your setup.
The three GPUs, and what every plan includes
NVIDIA L4 · 24 GB
Ada Lovelace, 300 GB/s, 72 watts, with hardware video encoders and decoders, AV1 included. For efficient inference, speech and video; one, two or four per server.
NVIDIA RTX PRO 6000 · 96 GB
Blackwell, with error-correcting GDDR7 memory and ray-tracing cores. For rendering, 3D and simulation, and models too large for an L4.
NVIDIA H200 · 141 GB
Hopper, with HBM3e memory at 4.8 TB/s — the most memory and bandwidth here, for large models, long context and fine-tuning.
Storage and transfer
100 GB of NVMe storage, 10 TB of transfer a month and one IPv4 address on every plan. More storage: price on request.
Ready for your framework
Drivers, CUDA and PyTorch installed on request, at no extra charge. Root access, so you can install anything else.
A GPU of your own
The whole GPU is assigned to your server, and no one else's work runs on it.
GPUs in India, billed in India
GPU servers run in our Mumbai data centre, so the data your models read — customer records, documents, recordings — stays in India, and users across India get quick responses.
- Priced in INR, exclusive of 18% GST · GST invoice from a Jaipur-registered business
- Support in English and Hindi, in IST
- Certifications held by the facility and operator: SOC 2 Type II · PCI DSS · SSAE 16
Which GPU for which job
| If you run | Start with | Why |
|---|---|---|
| A chatbot or assistant on a 7–8B open model | L4 ×1 | 24 GB holds the model at 16-bit, with room for context |
| Speech-to-text, OCR or video analytics | L4 | Built for efficient inference and video |
| Several models side by side, or more users | L4 ×2 or ×4 | One model per GPU, or share the traffic |
| A 70B model at 8-bit, or rendering and 3D | RTX PRO 6000 | 96 GB of memory on one card |
| The largest open models, long context, fine-tuning | H200 | 141 GB and the fastest memory here |
| Websites, databases or an ERP | Dedicated servers | No GPU needed |
Moving from another GPU provider, or from AWS or Azure? We set up the new server, copy your models and data, and test them on the new GPU before you switch. How migration works
Questions people ask
What is GPU cloud?
A server with one or more graphics processors (GPUs) that you rent rather than buy. GPUs run AI models, rendering and video far faster than ordinary processors. You get root access to a server in Mumbai with the GPU attached, and pay monthly.
How much does a GPU server cost per hour in India?
Our plans are monthly. One L4 at ₹37,499/month works out to about ₹51 an hour if it runs all month (730 hours), and the H200 to about ₹383. Hourly billing for short jobs: on request.
Which GPU should I choose: L4, RTX PRO 6000 or H200?
Choose by memory first. The L4 (24 GB) suits inference on models up to about 8 billion parameters, speech and video. The RTX PRO 6000 (96 GB) takes models up to about 70 billion parameters at 8-bit, plus rendering and 3D. The H200 (141 GB) is for the largest models, long context and fine-tuning.
How much GPU memory do I need for an LLM?
To run a model, allow about 2 GB per billion parameters at 16-bit, roughly half that at 8-bit and a quarter at 4-bit — plus room for context, which grows with the length of the conversation. A 7B model needs about 14 GB at 16-bit; a 70B model about 70 GB at 8-bit. Full fine-tuning needs roughly nine times as much; LoRA and QLoRA need far less.
Is the GPU shared with anyone?
The whole GPU is assigned to your server, and no one else's work runs on it.
Do you install CUDA and PyTorch?
Yes, on request — the GPU drivers, CUDA and PyTorch, at no extra charge. You have root access, so you can add anything else: vLLM, Ollama, Triton or your own stack.
How quickly can I get a GPU server?
Usually the same working day, when the GPU is in stock. GPUs sell out, so we confirm the one you want is available before we send the invoice.
Is my data kept in India?
Yes. GPU servers run in Mumbai, so your models and data stay in India.
Can I get a GST invoice and pay in rupees?
Yes. Prices are in rupees, exclusive of 18% GST, and you get a GST invoice from Chandnas, a Jaipur-registered business. The price stays the same at renewal.
Can I start small and move to a bigger GPU?
Yes. Start on one L4, then move to two or four, an RTX PRO 6000 or an H200 when you need more memory. We move your environment across.
Tell us the model. We'll size the GPU.
Model, users and hours a day — one recommendation, with the price.
Support 9 am – 7 pm IST. Server-down issues are answered 24/7, in English and Hindi.
Tell us what you run and we'll reply with one recommendation.
No sequence, no sales calls. One reply from an engineer, within working hours.