Token-billed models include an approximate price per image, video second, or LLM request
above their token rates. Hover over an estimate for its assumptions. Estimates use
current customer prices, including applicable discounts. Image estimates cover output only; input and
text/reasoning charges may apply separately. Video estimates with video input are per
billable second, including both input and output duration. GPT Image examples assume
1,000 output tokens per image, not a fixed size or quality. LLM examples use 1,000 uncached
text input tokens and 1,000 output tokens; the 128K–256K context tier uses 160,000 input
tokens and 1,000 output tokens to fit that tier. Audio and cache storage are excluded.
Actual usage can vary; estimates do not change token-based billing.
Billing behavior
- Creating a generation task can reserve wallet balance while the task is running.
- Successful billable requests are recorded once using a stable request or task identifier.
- Failed or cancelled tasks are released or remain non-billable according to the endpoint lifecycle.
- Your Dashboard wallet and ledger show the final customer charge.
The list shows final customer prices only. Model pages describe capabilities, not a fixed
price snapshot.
