How Billing Works
Modelpilot uses a straightforward credit-based billing system to make costs transparent and predictable.
Credit System
- Credits: The basic unit of payment on Modelpilot
- Conversion: 1 credit = $1.00 USD
- Billing: Your account is charged in credits based on the compute resources you use
10-Minute Billing Increments
Unlike some platforms that bill by the second, Modelpilot uses 10-minute billing increments:
- Deployments are billed in whole 10-minute blocks
- Even if you use a deployment for less than 10 minutes, you're billed for the full 10-minute period
- No charges during startup: The billing clock starts only when your deployment reaches the "Running" state
- No charges for failed deployments: If your deployment fails to start, you're not charged
- Stopped deployments don't incur compute charges
Cost Optimization Tips
- Batch your work: Run multiple tasks within the same 10-minute billing increment
- Stop idle deployments: Remember to stop deployments when not in use
- Schedule deployments: For recurring tasks, schedule deployments to run only when needed
- Choose appropriate hardware: Don't over-provision for your workloads
- Use spot instances: For non-critical workloads, spot instances can save up to 70%
Instance Pricing
Modelpilot offers various instance types to meet different needs. Prices vary based on the hardware specifications.
CPU Instances
| Instance | Specifications | Price (per hour) | Price (per 10-min) |
|---|
| CPU Standard | 3 vCPU, 16GB RAM | $0.11/hour | $0.02/10-min |
GPU Instances
Prices include compute, storage, and service fees. Actual prices are shown during deployment configuration.
| Instance | Specifications | Price (per hour) | Price (per 10-min) |
|---|
| GPU-RTX4090 | NVIDIA RTX 4090, 24GB VRAM, 6 vCPU | ~$0.79/hour | ~$0.13/10-min |
| GPU-L4 | NVIDIA L4, 24GB VRAM, 12 vCPU | ~$0.55/hour | ~$0.09/10-min |
| GPU-A6000 | NVIDIA A6000, 48GB VRAM, 9 vCPU | ~$1.47/hour | ~$0.25/10-min |
| GPU-A100-80GB | NVIDIA A100, 80GB VRAM, 8 vCPU | ~$2.86/hour | ~$0.48/10-min |
| GPU-H100 | NVIDIA H100, 80GB VRAM, 16 vCPU | ~$4.39/hour | ~$0.73/10-min |
Storage
Storage costs are included in the hourly rates shown above. Each deployment includes 200GB of persistent storage for models, data, and generated content.
Billing Examples
Example 1: Text Generation Model
- Model: Mistral 7B
- Instance: GPU-L4 ($0.55/hour)
- Usage Pattern: Deployed for 3 hours
- Billing: 18 x 10-minute increments = $1.65
Example 2: Video Generation Model
- Model: Wan 2.1
- Instance: GPU-A6000 ($1.47/hour)
- Usage Pattern: Deployed for 5 hours
- Billing: 30 x 10-minute increments = $7.35
Managing Your Billing
Adding Credits
- Navigate to the Billing page from your dashboard
- Click "Add Credits"
- Select the amount of credits you want to purchase
- Complete the payment process using your preferred method
Viewing Usage
- Go to your dashboard
- Click on the Billing tab
- View your current credit balance and usage history
- Download detailed usage reports if needed
Setting Spending Limits
To avoid unexpected charges, you can set spending limits:
- Navigate to the Billing page
- Click "Manage Spending Limits"
- Set a maximum amount of credits to be used per day/week/month
- Enable notifications for when you reach certain thresholds
FAQs
Q: Do I pay for deployments in "Stopped" state?
A: No, you only pay for running deployments. However, you will still be charged for any storage used.
Q: How do I estimate costs before deployment?
A: During the deployment process, the Review step shows estimated hourly and daily costs based on your configuration.
Q: Can I share credits across my organization?
A: Yes, with an Organization plan, credits can be shared among team members with appropriate permissions.
Q: What happens if I run out of credits?
A: Running deployments will be automatically stopped when your credit balance reaches zero.
Q: Can I get a refund for unused credits?
A: Credits are non-refundable, but they never expire as long as your account remains active.