Pricing
GPUs from ₹11/hour, images from ₹2
Rent a machine by the second, or skip the machine and pay per image (video from ₹42 a clip). Both billed in rupees, from the same wallet, at a fraction of what the big clouds charge for the same work.
Live prices for every available machine are on the marketplace — they’re the same rates we bill.
How billing works
Simple enough to explain in a line
Top up once, then spend it either way: start a machine and pay for the seconds it runs, or generate an image or a clip and pay for that. One wallet, no plans, no commitment.
Billed by the second
The meter starts when your instance is ready and stops the moment you stop it. No rounding up to the hour, no minimum session.
Prepaid, never a surprise bill
You top up a wallet and spend it down. You can't be charged more than you've loaded — a run simply stops when the balance is gone.
Priced in rupees
Pay by UPI, card or netbanking. No international card, no forex markup, no foreign transaction fee taking a bite out of every top-up.
One rate per GPU class
We publish the rate for each GPU type — hosts don't bid it up or down. What you see on a machine is what you pay.
What you’ll pay
It depends on the card you pick
Entry-level GPUs start at ₹11/hour — enough for inference, learning and small fine-tunes. Faster cards with more VRAM cost more per hour but finish the job sooner, so the cheapest card isn’t always the cheapest run. We publish a rate for each of 44 GPU classes, and you always see the exact hourly price on a machine before you start it.
Because our capacity comes from real machines rather than a datacenter full of hardware we had to buy, our rates sit well below traditional cloud providers for comparable GPUs.
For renters
What you pay, per GPU
One published rate per GPU class, billed by the second — hosts don’t bid it up or down. What you see on a machine is what you pay.
| GPU | VRAM | Disk | You pay |
|---|---|---|---|
| NVIDIA H100 | 80 GB | 20 GB | ₹208/hr |
| NVIDIA A100 80GB | 80 GB | 20 GB | ₹79/hr |
| NVIDIA A100 40GB | 40 GB | 20 GB | ₹71/hr |
| RTX 5090 | 32 GB | 20 GB | ₹50/hr |
| RTX 4090 | 24 GB | 20 GB | ₹47/hr |
| RTX 5080 | 16 GB | 20 GB | ₹43/hr |
| RTX 3090 Ti | 24 GB | 20 GB | ₹41/hr |
| RTX 4080 | 16 GB | 20 GB | ₹41/hr |
| RTX 5090 Laptop | 24 GB | 20 GB | ₹36/hr |
| RTX 5070 Ti | 16 GB | 20 GB | ₹36/hr |
| RTX 4070 Ti SUPER | 16 GB | 20 GB | ₹36/hr |
| RTX 5080 Laptop | 16 GB | 20 GB | ₹32/hr |
| RTX 4090 Laptop | 16 GB | 20 GB | ₹32/hr |
| RTX 4070 Ti | 12 GB | 20 GB | ₹32/hr |
| RTX 4070 SUPER | 12 GB | 20 GB | ₹32/hr |
| RTX 3090 | 24 GB | 20 GB | ₹29/hr |
| RTX 5070 | 12 GB | 20 GB | ₹28/hr |
| RTX 3080 Ti | 12 GB | 20 GB | ₹28/hr |
| RTX 5070 Ti Laptop | 12 GB | 20 GB | ₹25/hr |
| RTX 4080 Laptop | 12 GB | 20 GB | ₹25/hr |
| RTX 4070 | 12 GB | 20 GB | ₹25/hr |
| RTX 3080 | 10 GB | 20 GB | ₹25/hr |
| RTX 5060 Ti | 8 GB | 20 GB | ₹25/hr |
| RTX 3070 Ti | 8 GB | 20 GB | ₹23/hr |
| RTX 2080 Ti | 11 GB | 20 GB | ₹23/hr |
| RTX 3080 Ti Laptop | 16 GB | 20 GB | ₹21/hr |
| RTX 3070 | 8 GB | 20 GB | ₹21/hr |
| RTX 5060 | 8 GB | 20 GB | ₹21/hr |
| RTX 4060 Ti | 8 GB | 20 GB | ₹19/hr |
| RTX 2080 SUPER | 8 GB | 20 GB | ₹19/hr |
| RTX 3080 Laptop | 8 GB | 20 GB | ₹17/hr |
| RTX 3070 Ti Laptop | 8 GB | 20 GB | ₹17/hr |
| RTX 3060 Ti | 8 GB | 20 GB | ₹17/hr |
| RTX 2080 | 8 GB | 20 GB | ₹17/hr |
| RTX 5070 Laptop | 8 GB | 20 GB | ₹16/hr |
| RTX 4070 Laptop | 8 GB | 20 GB | ₹16/hr |
| RTX 4060 | 8 GB | 20 GB | ₹16/hr |
| RTX 2070 | 8 GB | 20 GB | ₹16/hr |
| RTX 5060 Laptop | 8 GB | 20 GB | ₹14/hr |
| RTX 4060 Laptop | 8 GB | 20 GB | ₹14/hr |
| RTX 3070 Laptop | 8 GB | 20 GB | ₹14/hr |
| RTX 3060 | 12 GB | 20 GB | ₹14/hr |
| RTX 3060 Laptop | 6 GB | 20 GB | ₹11/hr |
| GTX 1080 Ti | 11 GB | 20 GB | ₹11/hr |
| Extra disk — added in 30 GB blocks | +30 GB | +₹1.25/hr | |
Billed per second from your prepaid wallet — no setup fees, egress or minimums. An unlisted GPU is priced automatically from its measured compute and VRAM. Disk is billed the same way, and a listing’s price already includes all the disk that host shares — pick a smaller size at launch and it comes down.
Instance disk is temporary — your workspace is erased when the machine stops, which is also what keeps your work off the host’s computer afterwards. Download anything you want to keep before you stop it.
Don’t want a machine?
Pay per image or video instead
Type a prompt and pay for the picture or the clip — no GPU to pick, nothing to install, and no hourly meter running while you think. Same prices in the app and through the API. Same wallet, same hosts.
| What you get | Typically | Price |
|---|---|---|
| Images | ||
| FastQuick and affordable. | ~2s | ₹2 |
| StandardGreat quality and speed. | ~5s | ₹3 |
| PremiumMaximum detail. Takes a little longer. | ~15s | ₹5 |
| UpscaleMake an image you already generated bigger and sharper. | ~4s | ₹2 |
| Video | ||
| Quick · 3s clipShort clip · 3 seconds. | ~15 min | ₹42 |
| Standard · 5s clipBetter motion · 5 seconds. | ~23 min | ₹76 |
| SharpenMake a clip you already generated twice the size and sharper. | ~3 min | ₹17 |
Charged only when you actually get your image or clip. A generation that fails costs nothing — we retry it on another GPU at our expense. A job still waiting for a GPU can be cancelled for free.
Which is cheaper? Under roughly 15–30 images an hour, paying per image wins. Above that, rent the GPU from the table above and generate as much as you like.
For hosts
What you earn, per GPU
You keep 80% of every rupee a renter pays — GHive’s only fee is the 20% in between. No listing fee, no payout fee. You’re paid via UPI.
| GPU | VRAM | Disk | You earn |
|---|---|---|---|
| NVIDIA H100 | 80 GB | 20 GB | ₹166.40/hr |
| NVIDIA A100 80GB | 80 GB | 20 GB | ₹63.20/hr |
| NVIDIA A100 40GB | 40 GB | 20 GB | ₹56.80/hr |
| RTX 5090 | 32 GB | 20 GB | ₹40.00/hr |
| RTX 4090 | 24 GB | 20 GB | ₹37.60/hr |
| RTX 5080 | 16 GB | 20 GB | ₹34.40/hr |
| RTX 3090 Ti | 24 GB | 20 GB | ₹32.80/hr |
| RTX 4080 | 16 GB | 20 GB | ₹32.80/hr |
| RTX 5090 Laptop | 24 GB | 20 GB | ₹28.80/hr |
| RTX 5070 Ti | 16 GB | 20 GB | ₹28.80/hr |
| RTX 4070 Ti SUPER | 16 GB | 20 GB | ₹28.80/hr |
| RTX 5080 Laptop | 16 GB | 20 GB | ₹25.60/hr |
| RTX 4090 Laptop | 16 GB | 20 GB | ₹25.60/hr |
| RTX 4070 Ti | 12 GB | 20 GB | ₹25.60/hr |
| RTX 4070 SUPER | 12 GB | 20 GB | ₹25.60/hr |
| RTX 3090 | 24 GB | 20 GB | ₹23.20/hr |
| RTX 5070 | 12 GB | 20 GB | ₹22.40/hr |
| RTX 3080 Ti | 12 GB | 20 GB | ₹22.40/hr |
| RTX 5070 Ti Laptop | 12 GB | 20 GB | ₹20.00/hr |
| RTX 4080 Laptop | 12 GB | 20 GB | ₹20.00/hr |
| RTX 4070 | 12 GB | 20 GB | ₹20.00/hr |
| RTX 3080 | 10 GB | 20 GB | ₹20.00/hr |
| RTX 5060 Ti | 8 GB | 20 GB | ₹20.00/hr |
| RTX 3070 Ti | 8 GB | 20 GB | ₹18.40/hr |
| RTX 2080 Ti | 11 GB | 20 GB | ₹18.40/hr |
| RTX 3080 Ti Laptop | 16 GB | 20 GB | ₹16.80/hr |
| RTX 3070 | 8 GB | 20 GB | ₹16.80/hr |
| RTX 5060 | 8 GB | 20 GB | ₹16.80/hr |
| RTX 4060 Ti | 8 GB | 20 GB | ₹15.20/hr |
| RTX 2080 SUPER | 8 GB | 20 GB | ₹15.20/hr |
| RTX 3080 Laptop | 8 GB | 20 GB | ₹13.60/hr |
| RTX 3070 Ti Laptop | 8 GB | 20 GB | ₹13.60/hr |
| RTX 3060 Ti | 8 GB | 20 GB | ₹13.60/hr |
| RTX 2080 | 8 GB | 20 GB | ₹13.60/hr |
| RTX 5070 Laptop | 8 GB | 20 GB | ₹12.80/hr |
| RTX 4070 Laptop | 8 GB | 20 GB | ₹12.80/hr |
| RTX 4060 | 8 GB | 20 GB | ₹12.80/hr |
| RTX 2070 | 8 GB | 20 GB | ₹12.80/hr |
| RTX 5060 Laptop | 8 GB | 20 GB | ₹11.20/hr |
| RTX 4060 Laptop | 8 GB | 20 GB | ₹11.20/hr |
| RTX 3070 Laptop | 8 GB | 20 GB | ₹11.20/hr |
| RTX 3060 | 12 GB | 20 GB | ₹11.20/hr |
| RTX 3060 Laptop | 6 GB | 20 GB | ₹8.80/hr |
| GTX 1080 Ti | 11 GB | 20 GB | ₹8.80/hr |
| Extra disk you share — per 30 GB block | +30 GB | +₹1.00/hr | |
Earnings accrue by the second while your GPU is actually rented (and only then). You share 20 GB by default and can change it any time without reinstalling — share only what you can spare, since a rental that outgrows your offer is stopped to protect your drive. Set it up on the host page.
What’s included
- 20 GB of disk included with every rental
- No setup or platform fees
- No egress or bandwidth charges
- No minimum spend or commitment
- Ready-made PyTorch, Stable Diffusion & LLM environments
- Failed image and video generations are never charged
- Itemised transaction history