--- title: Billing --- # Billing E2E TIR offers flexible pricing models designed to support both short-term experimentation and long-term production workloads. You can choose between **On-Demand (Hourly)** and **Committed** billing based on your usage pattern and cost strategy. --- ## 1. On-Demand Pricing On-Demand pricing lets you pay only for the time your instance is actively running. Billing is calculated per minute, so you are charged only for actual usage, making it ideal for short-term workloads, testing, and variable usage patterns. ### How Billing Works | Event | Billing Status | |-------|---------------| | Instance enters **Running** state | Billing starts | | Instance enters **Stopped** state | Billing pauses | | Instance is **restarted** | Billing resumes | | Instance is **deleted** | Billing stops permanently | > You are not billed while the instance is in a **Creating** state. ### Example If the hourly rate is ₹60/hr and you run the instance for 5 minutes: ```bash Hourly Rate : ₹60/hr Usage : 5 minutes ₹60 ÷ 60 = ₹1 per minute ₹1 × 5 = ₹5 total billed ``` You are charged only ₹5 for 5 minutes of usage, rather than a full hour. --- ## 2. Committed Billing Committed billing allows you to reserve an instance for a fixed duration. The full amount is charged upfront at the time of instance creation, and in return you get a lower rate compared to On-Demand pricing. This model is best suited for long-running, predictable workloads where consistent compute availability is essential. > Before launching a committed instance, ensure your account has sufficient credits to cover the full committed price. The total amount is deducted immediately upon creation. ### Available Commitment Periods Available commitment periods: **30 Days**, **90 Days**, and **365 Days**. > The longer the commitment period, the lower the effective rate compared to On-Demand billing. ### How Billing Works | Event | Billing Status | |-------|---------------| | Instance is **created** | Full committed amount deducted upfront | | Instance is **running** | No additional charges during the period | | Instance is **deleted before period ends** | No refund for the remaining period | ### Example If you launch a 30-day committed instance on **June 5, 2026**, the full amount is deducted immediately and your instance runs uninterrupted until **July 5, 2026** with no additional compute charges. ```bash Committed Plan : 30 Days Start Date : June 5, 2026 End Date : July 5, 2026 Amount Deducted : ₹2,500 (upfront, at launch) Additional Cost : ₹0 during the 30-day period ``` At the end of the period, your selected post-commitment policy takes effect automatically. --- ## 3. Converting On-Demand to Committed If you started with an On-Demand instance and your workload has stabilized, you can convert it to a Committed plan to benefit from lower long-term rates without creating a new instance. ### Steps to Convert 1. Navigate to **Instances** and select your On-Demand instance. 2. Stop the instance. 3. Click **Actions** and select **Convert to Committed**. 4. Select your desired commitment period and post-commitment policy. 5. Review and confirm the commitment terms. > Ensure your account has sufficient credits to cover the full committed amount before proceeding. The amount is deducted immediately upon conversion. --- ## 4. Upgrading On-Demand Configuration On-Demand instances offer full flexibility to change your compute configuration at any time. You can scale your resources up or down, or switch between instance types entirely, based on your evolving workload requirements. After the plan upgrade, billing will be calculated as per the new configuration. ### Supported Configuration Changes | Change Type | Example | |-------------|---------| | **Scale Up** | Move from 1 GPU to 4 GPUs as training workload grows | | **Scale Down** | Move from 4 GPUs to 1 GPU during lighter workloads | | **Switch to GPU** | Move from a CPU instance to a GPU instance for model training | | **Switch to CPU** | Move from a GPU instance to a CPU instance for data preprocessing | > Configuration changes are available only for **On-Demand instances**. Committed instances cannot be upgraded, downgraded, or switched to a different instance type during the commitment period. ### How to Update Configuration 1. Navigate to **Instances** and select your instance. 2. Click **Actions** and select **Update Plan**. 3. Choose your desired instance type and configuration. 4. Confirm the update. > It is strongly recommended to save your image before updating the plan to avoid any environment or data loss. Refer to the [Save Image](#) section for steps. --- ## 5. Post-Commitment Options When creating a committed instance, you must select a post-commitment policy that defines what happens automatically when the commitment period ends. This must be selected at the time of instance creation. ### Auto Renewal *(Default)* The instance automatically renews for another commitment period using the same or a newly selected plan. If you want to switch to a different plan, you can select it before the current period ends and it will take effect once the commitment completes. ### Switch to Hourly Billing Once the commitment period ends, the instance automatically transitions to On-Demand hourly billing. No action is required from your side. ### Auto Deletion The instance is automatically terminated and deleted once the commitment period ends. Use this option when you know your workload has a defined end date. > Committed instances cannot be stopped during the commitment period. They can only be deleted. **No refunds are issued for the remaining commitment period upon deletion.** Review your decision carefully before deleting a committed instance. --- ## 6. Storage and Dataset Billing Storage resources attached to your instance are billed independently and are not affected by the instance lifecycle. This means storage charges continue even if your instance is stopped or deleted. Charges stop only when the storage resource itself is explicitly deleted. | Resource | Billed Independently | |----------|---------------------| | Shared File System (SFS) | Yes | | Parallel File System (PFS) | Yes | | Datasets | Yes | > Always review and delete unused storage resources to avoid unexpected charges after instance deletion. --- ## 7. Container Registry Billing Container Registry is billed independently and is not affected by the instance lifecycle. Charges continue even if your instance is stopped or deleted, and stop only when the Container Registry itself is explicitly deleted. --- ## 8. What is Not Billed The following actions and features are available at no additional charge on your instance. | Action | Details | |--------|---------| | **Add SSH Key** | Adding or updating SSH keys on your instance is free | | **Security Group** | Creating and attaching security groups incurs no charge | | **VPC IP** | Attaching a VPC IP to your instance is free; VPC is billed separately under My Account | | **Update Image** | Switching your instance image does not incur any additional cost | | **Launch App** | Opening and using the JupyterLab workspace is free | | **Start Script** | Creating and running start scripts on boot is free | --- ## Cost Estimation Use the [E2E Pricing Calculator](https://calculator.e2enetworks.com/) to compare On-Demand and Committed plans, estimate projected costs, and evaluate potential savings based on your workload duration and usage patterns before committing to a plan. ---