Billing for Compute

Krutrim Cloud offers transparent and usage-based pricing for all compute services, including VMs, AI Pods, and GPU Baremetals. This guide will help you understand how billing is calculated and what to expect on your invoices when using our compute infrastructure.

Billing Overview

Compute resources are billed based on actual usage, with rates displayed on an hourly basis. However, Krutrim applies fine-grained metering, ensuring you're only charged for what you use, in 15-minute billing intervals.

You are only billed while your resources are in an active state (running or initialising). Stopped or terminated resources do not incur compute charges, although associated storage or IPs may continue to incur costs.

Metering and Billing Logic

Even though prices are shown as per-hour rates, the actual metering happens in 15-minute increments. The minimum billable unit is 15 minutes.

Example Billing Scenarios:

Usage Duration
Billed Duration

1 hour 40 minutes

1 hour 45 minutes

2 hours 5 minutes

2 hours 15 minutes

35 minutes

45 minutes

14 minutes

15 minutes (minimum)

This granularity ensures that you are billed fairly and precisely, with charges rounded up to the nearest 15-minute mark.

Taking the 1st case as an example, if a user uses an A100-40GB-NVLINK-1x which is priced at Rs.105/hour and the user uses the VM for 1 hour 40 minutes, the billed duration will be 1 hour 45 minutes, hence the final bill will be 105 + 105*(45/60) = Rs. 183.75

Billing by Compute Type

Virtual Machines (VMs)

VMs are billed based on their CPU and GPU configurations (vCPU count, memory, and GPU type). Pricing varies by VM flavour and region.

Flavour
vCPU Count
RAM
Cost (Rs/Hour)

CPU-1x-4GB

1
4
3

CPU-2x-8GB

2
8
6

CPU-4x-16GB

4
16
13

CPU-8x-32GB

8
32
25

CPU-16x-64GB

16
64
49

CPU-32x-128GB

32
128
97

GPU Baremetals

GPU Baremetal servers provide exclusive access to full GPUs for high-performance workloads. These are custom-configured deployments and require you to contact our sales team for pricing and provisioning.

To request GPU Baremetal pricing, please reach out to our Sales team at [email protected]

Billing

Our Billing cycle happens every 15 minutes, but you will be charged for the duration during which you used the pod during those 15 minutes.

Example: Let’s say you started a pod at 10:00 and you terminated the pod at 10:10. ➡️ You will only be charged for 10 minutes.

Other Considerations

  • Attached Storage: Any block storage volumes attached to compute instances will be billed separately based on the size and duration of attachment.

  • Elastic IPs: IPs assigned to stopped or unattached VMs may incur additional costs.

  • Snapshots and Backups: Charged independently based on size and retention period.

  • Overage or Quota Breach: Exceeding free-tier or reserved quota (if applicable) will automatically incur standard pricing.

Where to View Usage

You can track your ongoing usage and estimated billing:

  • In the Billing Dashboard on the Krutrim Console

  • Via API or SDK for programmatic access to usage summaries

  • Through alerts and usage reports, which can be configured by account or project

Last updated

Was this helpful?