Every AI feature is priced in tokens: video, images, audio, captions and upscaling alike. The cost is shown before anything runs, so nothing is a surprise.
Where tokens come from Paid plans include a monthly token allowance that renews with your billing cycle; one-off top-up packs add more whenever you need them. Your balance lives in the top bar.
How prices are set Each model's compute cost drives its token price, scaled by your settings: a 10-second 1080p video costs more than a 5-second 480p one. The price shown at the button is the price charged.
Confirm before spending Costed actions (generation, captions, upscales) display their price up front; captions add an explicit confirm step with the audio length being billed.
Automatic refunds If a run fails or returns nothing usable (a failed render, captions with no speech), the charge is refunded to your balance automatically.
Track spending The billing page shows your allowance, your top-ups and a full transaction ledger, so every token in and out is accounted for.
Tips
Trim first, transcribe second: captions bill only the audio on your timeline.
Variants are charged per variant, and two considered candidates usually beat four random ones.
Editing, exporting and everything that is not AI is free. Tokens only ever pay for AI compute.