Pricing
Published.
All of it.
Every rate we charge for on-demand is published transparently. For 12-month reservations and 36-month commitments, we structure custom agreements tailored to your workload.
GPU · per GPU-hour
Accelerators
| Accelerator | Memory | Rate | Monthly equivalent |
|---|---|---|---|
| H200 SXM | 141 GB | ₹319 /hr | ≈ ₹2,32,870 /mo |
| H100 SXM | 80 GB | ₹239 /hr | ≈ ₹1,74,470 /mo |
| H100 PCIe | 80 GB | ₹199 /hr | ≈ ₹1,45,270 /mo |
| MI300X | 192 GB | ₹269 /hr | ≈ ₹1,96,370 /mo |
| A100 SXM | 80 GB | ₹149 /hr | ≈ ₹1,08,770 /mo |
| L40S | 48 GB | ₹99.0 /hr | ≈ ₹72,270 /mo |
| RTX 6000 Ada | 48 GB | ₹79.0 /hr | ≈ ₹57,670 /mo |
| RTX 4090 | 24 GB | ₹44.0 /hr | ≈ ₹32,120 /mo |
On-demand · billed per second · no minimum term
Compute
CPU and memory
| Resource | Rate |
|---|---|
| vCPU: C family | ₹1.10 /hr |
| vCPU: M family | ₹1.35 /hr |
| vCPU: R family | ₹1.90 /hr |
| vCPU: I family | ₹2.40 /hr |
| Memory | ₹0.55 /GB-hr |
| Public IPv4 | ₹180 /month |
Storage
Capacity
| Tier | Rate |
|---|---|
| Block NVMe | ₹8.00 /GB-mo |
| Shared filesystem | ₹6.20 /GB-mo |
| Object Standard | ₹1.60 /GB-mo |
| Object Archive | ₹0.40 /GB-mo |
| Snapshots | ₹1.60 /GB-mo |
| Egress · requests | ₹0.00 |
All rates exclusive of GST. Billing is per second for compute and GPU, per GB-month prorated daily for storage.
Included at every size
Not an upsell
Data egress
Unmetered, in and out, across every service.
Internal traffic
Free between compute, GPU and storage in a region.
24×7 support
Humans on chat, phone and email. No paid tier.
DDoS protection
Always on at the edge, no scrubbing surcharge.
Questions we get asked
Billing, plainly
Is there really no egress charge?
Correct. We don't meter data leaving our network, on any service. Our transit is a fixed cost we already carry, and metering it is a commercial choice rather than a physical one. If you push sustained multi-gigabit traffic we'll talk about a dedicated port, but there's no per-gigabyte line on the invoice.
How does per-second billing actually work?
A pod or instance accrues charges from the moment it reaches a running state to the moment it stops. Stopped instances are billed for attached storage only. There's no rounding to the hour and no minimum runtime, so a four-minute test costs four minutes.
What's the difference between reserved and committed?
Reserved holds specific capacity for you at a discounted rate for twelve months, billed whether or not you use it. Committed is a thirty-six month spend commitment across any service, with the deepest rates and flexibility to move between GPU types as your workload changes.
Do you accept GeM and standard enterprise procurement?
Yes. We're listed on GeM for public-sector buyers, and we work with the usual enterprise instruments: POs, monthly consolidated invoicing in INR, credit terms and MSA-based contracting. Startups can also pay by card with no contract at all.
Can I get credits to evaluate?
Yes. Every new account gets evaluation credits, and if you're benchmarking against another provider we'll usually fund a proper comparison run rather than a token trial. Ask for it directly.
Need custom capacity
or enterprise terms?
For multi-node clusters, annual reservations, or 36-month spend commitments, speak directly with our infrastructure team.