INR billing - OpenAI-compatible model API - GPU rental India
Rent GPUs Online
in India — live stock, INR billing
Lumino AI gives Indian developers GPU cloud access and OpenAI-compatible hosted model APIs from one account. Call chat, coding, video, and speech models, rent H100s and A100s by the hour, and pay in INR.
Browse live GPUs Contact sales
Model API - Chat completions - pay per token
GPU Cloud - H100 - A100 - RTX 4090
Video AI - Pixverse V6 generation
Compare GPU rental options
Start with memory, then compare the GPU rate. These are individual GPU configurations; available models and stock can change between visits.
GPU / hour · INR
GPU rates are shown per hour. Your configuration also includes storage charges. The marketplace shows the current price and availability before you rent.
Choose your path
Start with the infrastructure you need
Rent a full GPU for complete control, or use hosted models when you only need API output.
Model API
Call hosted models through one API
Generate an API key and call chat, coding, video, and speech models through OpenAI-compatible endpoints with INR billing.
- OpenAI-compatible chat completions
- Streaming and tool calling support
- Scoped API keys
- Pay for actual usage
GPU Cloud
Rent GPUs on demand
Inspect live inventory, rent H100, A100, RTX 4090 and more, and keep billing visible in INR.
- Live inventory before sign-up
- 60-second billing intervals
- SSH and browser terminal access
- No long-term contract
Clusters
Scale to multi-node training
Request 2 to 32 node H100 clusters with high-speed interconnect for large training runs.
- 2 to 32 node configurations
- High-speed interconnect
- Slurm and Ray ready
- Approval-based access
How renting works
From comparison to running pod in minutes
01
Compare live GPUs
Browse the marketplace without an account. Filter by VRAM, availability, and hourly INR rate. Every listing shows current stock before you pay anything.
02
Rent in minutes
Pick storage, choose a Docker image, and deploy. Connect over SSH or the browser terminal. Top up credits with UPI, cards, or netbanking.
03
Pay per second, stop anytime
Billing ticks per second while the pod runs, measured every 60 seconds. Stopping releases the GPU and stops compute billing — only retained storage keeps a small charge.
FAQ
GPU rental questions, answered
How much does GPU rental cost in India?
It depends on the GPU and live demand. Entry GPUs start under ₹30/hr while H100-class cards are premium tier. You pay per second of active compute in INR — check the marketplace for live prices before renting.
What happens when I stop my GPU?
Stopping releases the GPU back to the provider pool, so restarting later needs live capacity of the same GPU type. Your stopped pod and its volume data stay stored for a small storage charge until you terminate.
How do I choose between H100, A100, and RTX 4090?
Start with VRAM: fit model weights plus workspace in memory first, then compare throughput. The H100 is fastest, the A100 is the balanced workhorse, and the RTX 4090 wins on value for 24 GB workloads.
Do I need an account to see GPU prices?
No. Live inventory and hourly INR rates are public on the marketplace. You only need an account — and credits — when you actually rent.
How is GPU billing calculated?
Compute is billed per second at the displayed hourly rate, deducted in 60-second cycles. Stopping the pod stops compute billing. Persistent storage bills separately until terminated.
Get started
Start with the part you actually need
Use model APIs for AI outputs. Rent GPUs if you need compute. Keep the stack simple.
Try Models Rent a GPU