Generation pricing and billing

Use the current model catalog for rates, and your account or API key's quote for the settings you intend to submit. This guide explains billing conditions without publishing a second, stale price table.

View live model prices

Understand the billing unit

Per image or task: check whether the quote covers one result, a task, or the requested result count.
Per second: the model's published unit and applicable output or reference-duration rule determine the amount.
Per million tokens: the pre-submission amount is an estimate. Final settlement uses the actual token usage returned for the completed task.

Settings and permissions matter

Resolution, duration, reference-video use and result count can change the quote. Compare like-for-like settings, and confirm that your selected key has access to the model.

Public catalog information is not a promise of access for every account. Available groups, key scopes and applicable account pricing are resolved by the existing service. Use the same authorized key when reading a video quote and submitting the request.

Do not treat an estimate as a spending cap

Review the current amount and billing mode immediately before submission. An estimate for token usage is not a guaranteed fixed final charge. Do not assume a saved screenshot or previously downloaded guide is a live quote.

Model prices and availability can change as the catalog updates. Review any offer's current terms separately; account creation and generation submission are distinct actions.

If a request fails

Inspect the task's specific failure reason and its recorded charge or refund status. Check the wallet entry for a confirmed refund rather than inferring one from an error message.

Correct invalid material or settings before deciding whether to submit again. After an ambiguous timeout, check for an existing task first so you do not create a duplicate paid request.