Skip to main content

Per Second Billing

Our billing system provides precise, per-second billing accuracy while using hourly intervals for reporting and API queries. This guide explains how we calculate costs with second-level precision and what this means for your billing.

How Per-Second Billing Works

Precise Time Tracking

  • Start/Stop Times: We track the exact second when resources start and stop

  • Duration Calculation: Billing duration is calculated to the precise second

  • Fractional Hours: Costs are calculated using fractional hours (e.g., 4.504166667 hours = 4 hours 30 minutes 15 seconds)

Example: GPUaaS Pool Usage

Start Time: 2025-10-13 08:30:00
End Time:   2025-10-13 13:00:00
Duration:   4 hours 30 minutes
Billed As:  4.5 hours

📊 Understanding Fractional Hours

Common Fractional Hour Values

Duration

Fractional Hours

What You'll See in Billing

30 minutes

0.5 hours

"pool_hours": 0.5

1 hour 15 minutes

1.25 hours

"pool_hours": 1.25

2 hours 45 minutes

2.75 hours

"pool_hours": 2.75

4 hours 30 minutes

4.5 hours

"pool_hours": 4.5

8 hours 20 minutes

8.33 hours

"pool_hours": 8.33

Precision Examples

json
{
  "pool_hours": 4.583333,     // 4 hours 35 minutes
  "interval_cost": "38.25",   // $8.34/hour × 4.583333 hours
  "subscription_rate": "8.34"
}

Cost Calculation Examples

Basic Calculation

Hourly Rate: $8.34/hour
Usage Time: 2 hours 30 minutes (2.5 hours)
Cost: $8.34 × 2.5 = $20.85

Real-World Scenarios

Scenario 1: Short Usage

GPU Pool Usage: 45 minutes
Calculation: $8.34 × 0.75 hours = $6.26
You Pay: $6.26 (not a full hour)

Scenario 2: Partial Day

Instance Running: 6 hours 20 minutes
Calculation: $2.50 × 6.33 hours = $15.83
You Pay: $15.83

Scenario 3: Month-End Cleanup

Storage Mounted: 28 days 14 hours 30 minutes
Daily Breakdown:
- 28 full days: 28 × 24 = 672 hours
- Partial day: 14.5 hours
- Total: 686.5 hours
Cost: $0.10 × 686.5 = $68.65

📈 Billing Intervals vs. Billing Precision

Important Distinction

Aspect

How It Works

Billing Precision

Per-second accuracy - costs calculated to the exact second

API Query Intervals

Hourly minimum* - can query by hour, day, week, month

Cost Calculation

💰 Fractional hours - uses decimals for precise billing

*Hourly endpoints will be available from our 2.1 release.

What This Means

  • Users are billed precisely for the time used

  • API reports in hourly intervals for practical querying

  • No "rounding up" to full hours - users pay for exact usage

🔍 Reading Your Billing Data

API Response Format

json
{
  "2025-10-13": {
    "interval_hours": 4.583333,        // 4 hours 35 minutes
    "interval_cost": "38.25",          // Exact cost for 4.583333 hours
    "from": "2025-10-13T08:25:00Z",    // Precise start time
    "to": "2025-10-13T13:00:00Z",      // Precise end time
    "subscription_rate": {
      "price": "8.34"                  // Hourly rate
    }
  }
}

Key Fields to Understand

  • interval_hours: Exact usage time as fractional hours

  • interval_cost: Precise cost based on exact usage

  • from/to: Exact start and end timestamps

  • price: Hourly rate used for calculation

Practical Examples

Development Workflow

1. Start GPU instance: 09:15:30
2. Run training job: 2 hours 23 minutes
3. Stop instance: 11:38:30
4. Billed for: 2.383 hours (not 3 full hours)
5. Cost: $8.34 × 2.383 = $19.87

Testing Environment

1. Deploy test environment: 14:45:00
2. Run tests: 47 minutes
3. Cleanup: 15:32:00
4. Billed for: 0.783 hours
5. Cost: $5.00 × 0.783 = $3.92

📋 Frequently Asked Questions

Q: Why do I see decimal hours instead of minutes?

A: Our billing system uses industry-standard fractional hours (4.5 hours = 4 hours 30 minutes) for precise calculations while maintaining compatibility with hourly pricing models.

Q: How precise is the time tracking?

A: We track resource assignment to the second, then calculate costs using fractional hours for maximum accuracy.

Q: Can I query billing data by minute?

A: API queries are as granular as hourly, but the costs within each hour are calculated with per-minute precision.

Q: Why might I see values like 71.999999 hours?

A: This represents floating-point precision in calculations. 71.999999 hours is effectively 72 hours (3 days) in billing terms.