Who is it for?
- Cloud infrastructure — GPU/CPU reservations, storage commitments
- AI / ML platforms — inference-hour commitments, training-job minimums
- SaaS & API providers — API-call or data-transfer commitments
Key Terminology
How Billing Works
For each billing period, Flexprice evaluates one of two outcomes:- Usage ≥ Commitment → Total = Commitment + (Usage Cost − Commitment) × Overage Factor. The customer pays the committed amount at standard rate, plus a premium on everything above it.
- Usage < Commitment, True-Up enabled → Customer pays the full commitment amount (shortfall added as true-up fee).
- Usage < Commitment, True-Up disabled → Customer pays only for actual usage.
Example 1 — AI Inference Platform (Per-Period Commitment)
An AI platform sells inference compute at $2 / vCPU-hour. They want customers to commit to a minimum of 500 vCPU-hours/month and charge a 1.5× premium on any usage above that.Requirement
Pricing in the plan

Configure commitment from the line item
...) next to the vCPU-hours charge and select Configure Commitment.
Configure the commitment details

Scenario 1 — Usage exceeds commitment (700 vCPU-hours)
Scenario 2 — Usage below commitment, true-up enabled (300 vCPU-hours)
Scenario 3 — Usage below commitment, true-up disabled (300 vCPU-hours)
Example 2 — Cloud Infra Company (Windowed Commitment)
A cloud infrastructure company sells GPU compute at $2 / GPU-hour. Instead of a single monthly commitment, they want to commit customers on an hourly basis — a minimum of 10 GPU-hours every hour. Each hour settles independently and the totals are summed at the end of the billing period.Requirement
Pricing in the plan

Configure commitment from the line item
...) next to the GPU-hours charge and select Configure Commitment.
Configure the windowed commitment details

Hour 1 — Usage exceeds commitment (15 GPU-hours)
Hour 2 — Usage below commitment, true-up kicks in (6 GPU-hours)
Hour 3 — Usage matches commitment exactly (10 GPU-hours)
Billing period total

