PowerPod GPUsUse casesCase studiesGuidesBlogFor collegesHelp Sign inCreate account

GPU rental price in India per hour: today’s rates in rupees, updated daily

Price guide · Published 2 November 2026 · prices on this page are live

GPU prices in India (INR) · Nepal (NPR) · across Asia · everywhere else.

This page answers one question with live numbers: what does it cost to rent a GPU in India right now, per hour, in rupees? Every figure below is read from PowerPod’s board as you load the page. It is the lowest price of a verified host for that card, billed by the second, and the rupee column is what an account in India pays. The dollar and Nepali rupee columns are there for comparison.

Today’s hourly price, card by card

GPUMemoryUSDINRNPR
RTX 306012 GBUSD 0.09INR 9NPR 14
RTX 30708 GBUSD 0.14INR 13NPR 20
RTX 308010 GBUSD 0.20INR 19NPR 30
RTX 309024 GBUSD 0.30INR 28NPR 44
RTX 407012 GBUSD 0.16INR 15NPR 23
RTX 408016 GBUSD 0.39INR 37NPR 58
RTX 409024 GBUSD 0.60INR 56NPR 88
RTX A500024 GBUSD 0.37INR 35NPR 55
RTX A600048 GBUSD 0.65INR 61NPR 96
L40S45 GBUSD 0.86INR 80NPR 128
A100 PCIe40 GBUSD 0.65INR 61NPR 96
A100 SXM480 GBUSD 0.65INR 61NPR 96
H100 PCIe80 GBUSD 3.91INR 365NPR 580
H100 SXM80 GBUSD 3.74INR 349NPR 556
H200140 GBUSD 5.63INR 526NPR 836

Lowest price per hour today, billed by the second.

Three things to know about these prices. First, they move: hosts join and leave the market during the day, so the lowest price at noon can differ from the lowest price at midnight. Second, they are complete for GPU time: there is no separate charge for the machine around the card, and the only extras are disk space while the machine exists and data transfer, both shown before you rent. Third, the price you see when you press Rent is locked for that machine for as long as you keep it.

The cheapest card for each amount of memory

Most people do not need a particular card; they need a certain amount of GPU memory. This is the cheapest card on the board right now for each size:

MemoryCheapest GPU todayUSDINRNPR
8 GB or moreRTX 2060 12 GBUSD 0.08INR 8NPR 12
12 GB or moreRTX 2060 12 GBUSD 0.08INR 8NPR 12
16 GB or moreRTX 2080 Ti 22 GBUSD 0.13INR 12NPR 19
24 GB or moreTesla P40 24 GBUSD 0.18INR 17NPR 26
48 GB or moreQuadro RTX 8000 48 GBUSD 0.39INR 37NPR 58
80 GB or moreGB10 119 GBUSD 0.50INR 47NPR 74

A 12 GB card runs notebooks, classic machine learning, Whisper and Stable Diffusion 1.5. At 16 GB you can serve an 8B language model or fine-tune one with QLoRA. The 24 GB cards, the RTX 3090 and RTX 4090, are the sweet spot for fine-tuning and image work. At 48 GB you serve 70B-class models with 4-bit weights; the 80 GB cards are for speed and for the largest jobs. The guide to choosing a GPU goes through each job.

What a job costs in rupees

Hourly prices are abstract; jobs are not. These are typical runs on the cheapest suitable hosts today, with the rupee column first for accounts in India:

Fine-tune a 7B language model (LoRA, about 10,000 examples, 3 epochs)

GPUTypical timeUSDINRNPR
A100 SXM42 hUSD 1.30INR 122NPR 192
A100 PCIe2.2 hUSD 1.43INR 135NPR 212
RTX 30905 hUSD 1.50INR 140NPR 220
RTX 40903 hUSD 1.80INR 168NPR 264
RTX 50902.5 hUSD 1.88INR 175NPR 278

Generate 1,000 images (Stable Diffusion XL, 1024×1024, 30 steps)

GPUTypical timeUSDINRNPR
RTX 30902.4 hUSD 0.72INR 68NPR 106
RTX 40901.2 hUSD 0.72INR 68NPR 106
A100 SXM41.2 hUSD 0.78INR 74NPR 116
RTX 50901.1 hUSD 0.83INR 77NPR 123
RTX A50002.8 hUSD 1.04INR 98NPR 154

A working day in JupyterLab (8 hours)

GPUTypical timeUSDINRNPR
RTX 30608 hUSD 0.72INR 72NPR 112
RTX A40008 hUSD 1.12INR 104NPR 160
RTX 30908 hUSD 2.40INR 224NPR 352
RTX 40908 hUSD 4.80INR 448NPR 704
A100 SXM48 hUSD 5.20INR 488NPR 768

Notice how the big cards close the gap on jobs: an H100 costs several times an RTX 4090 per hour but finishes a fine-tune in a fraction of the time, so the bill for the job is much closer than the hourly rate suggests. For a one-off job, pick by total cost and your deadline, not by the hourly price.

How this compares with GPU clouds based in India

The Indian GPU clouds publish starting prices on their own pages; these were the figures in their page titles on 30 October 2026, and they may have changed since. PowerPod is not the cheapest for every card on every day. What differs is the model: an open market of verified hosts, prices in rupees, billing by the second with no plan or commitment, consumer cards such as the RTX 4090 next to data-centre cards, and software that is ready to run.

Paying from India

Accounts in India are billed in Indian rupees. You add credits by card on a secure Stripe page with 3-D Secure; the smallest top-up is NPR 200. The payment is processed abroad, so switch on international transactions for the card in your bank’s app first; most Indian banks keep that switch under card controls. Credits appear the moment the payment goes through, and a monthly statement in rupees is in the console. The India page has the full picture, including where today’s machines are and which are nearest.

Create accountSee all GPU prices

Questions

What is the cheapest GPU to rent in India today?

The cheapest card with at least 8 GB of memory on PowerPod right now is the RTX 2060 at USD 0.08 · INR 8 · NPR 12 an hour. The cheapest card is rarely the cheapest way to finish a job, though: see the job tables for what a fine-tune or a batch of images costs on each card.

Is the price per hour or per second?

The price is quoted per hour and billed per second: a machine that runs for 17 minutes costs 17 minutes. Stopping a machine ends GPU billing at once; only its disk is billed until you destroy it.

Are there hidden charges?

No. The hourly price covers the GPU and the machine around it. Disk space is billed per GB-month at the rate shown before you rent, and data transfer at the host’s per-GB price, also shown before you rent. There is no minimum term and no monthly fee.

Do the prices include GST?

Prices are shown as they will be charged to your account. PowerPod is not an Indian company and does not add Indian GST to the price; your bank may add its own charge for an international card payment.

More from the blog

Run Stable Diffusion on a cloud GPU from India: ComfyUI or Forge in minutes, paid in rupeesTutorialFine-tune Llama on an H100: what it costs in rupees, and when the big card is cheaperTutorialRun Ollama and Open WebUI on a rented GPU: your own ChatGPT for an eveningTutorialTrain your own SDXL LoRA with Kohya’s GUI on a rented GPU, in about an hourTutorialAll posts