Skip to main content

Documentation Index

Fetch the complete documentation index at: https://rendobar.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Rendobar bills every job in nanodollars (one-billionth of a US dollar) drawn from your organization’s prepaid balance. This page is the why; for the how-to (plans, balance, top-up), see Credits & billing.

​
The two billing models

Usage-metered APIs pick one:
  1. Tier-based quotas — you buy a plan, get a bucket of minutes or operations, overage bills flat. Easy to understand, hard to price varying workloads.
  2. Credit-based nanodollars — every job has a cost formula, the balance is prepaid, charges are per-job with sub-cent precision. Precise, composable, takes a sentence to explain.
We picked nanodollars.

​
Why

  • Cost matches resources used. A 30-second watermark on 1080p shouldn’t cost the same as a 10-minute H.265 transcode of a 4K master. Nine decimal places of precision means the bill matches the work.
  • Prepaid means predictable budgets. You top up when you want, jobs run only if affordable, no end-of-month surprise bill. The dashboard shows formatted dollars; the nanodollar abstraction stays internal.

​
What you give up

  • You think in dollars, not operations. “Can I run this?” becomes “Do I have enough credits?” — not “Do I have enough quota?”. Every job response shows both cost.nanodollars and cost.formatted.
  • Pricing can change. Per-job formulas mean we can adjust pricing when underlying infra changes. When that happens, it’s announced in the changelog 30 days ahead.

​
Credit lifecycle

  1. Pre-check. Submit triggers a balance check against the estimated cost. No deduction. Insufficient balance returns 402 INSUFFICIENT_CREDITS.
  2. Execution. Job runs. Balance untouched.
  3. Debit. On success, actual cost is computed (real compute time or output duration) and atomically deducted.
  4. Refund. Failure or cancellation never deducts. Failed jobs cost zero.

​
Cost models in flight

Today only one model is live: per-compute-time for raw.ffmpeg (~$0.05/min). As dedicated job types ship, they’ll add per-input-duration, per-output-duration, and flat-rate models — each documented on the job type’s reference page.

​
Practical advice

  • Start on Free. $5 signup grant is enough for many hundreds of small jobs.
  • Use rb_test_* keys while building. Test-mode runs the full pipeline, returns real output, doesn’t deduct credits.
  • Store cost.nanodollars from every response. Reconcile against your monthly invoice.

​
See also