Published tiers, quoted up front by file count or volume, no quote call and no metered per-document bill that only adds up after the fact.
What metered API billing actually feels like
Cloud Document AI, transcription, and translation APIs almost universally bill by usage — per page, per minute, per character, per API call. It's cheap to start and genuinely difficult to predict at scale: a batch job that looked small on your desk can turn into a bill you didn't expect, because nobody budgets in units of API calls.
The trade a flat job price makes
A job is priced by a published tier before you submit it — you know the number before you commit, not after a bill arrives. The trade-off runs the other way too: a tier is a range, so a job just over a boundary pays the next tier up, where a metered API would have charged you only for the extra unit.
Where this breaks down — said plainly
- Tiered pricing isn't always cheaper than metered billing — for very small occasional jobs, a pay-per-call API can come out lower. The appeal here is predictability, not a guarantee of the lowest possible price every time.
- Rush pricing is a separate, explicit add-on where available — not folded invisibly into the base tier.
Related guides
Where this fits on the site: this is part of how every job runs here, across all services. See Services & Pricing for what's actually sold, or go straight to Choose your job.