Pay by actual usage, with costs visible.
Training is measured by GPU time and resource configuration. Inference is measured by requests or tokens. App and compute-sharing usage flows into one ledger so teams can review cost and revenue records.
This page explains metering, budget, and revenue-record rules only; available workflows follow what is opened in the console.
Training records cost by actual GPU time and resource configuration. Owned, rented, and platform-managed machines use the same metering rules.
Inference records usage by request, token, or app price. Consumer spend, platform service fees, and provider revenue remain traceable in the ledger.
Idle machines can stay private or be shared to the compute marketplace under your rules. AionRay tracks rental time, usage outcomes, and revenue records for review.
Metering rules
AionRay does not require a fixed subscription tier first. Start by connecting machines, deploying services, or creating training jobs, then review the ledger by actual usage.
| Item | Current rule |
|---|---|
| Organization onboarding | Enabled by default, no plan selection first |
| Platform billing model | Measured across training, inference, apps, and rentals |
| Budget controls | Console records balance, budget, and usage |
| Training | By GPU time and resource configuration |
| Inference / App services | By request, token, or app price |
| Compute-sharing revenue | By rental time, usage outcomes, and revenue records |