Skip to main content
API runs draw from the same credit balance the visual editor uses. Every run is billed in credits; the exact amount appears on the run record as credits_charged once it completes.

What you pay for

API runs include a small premium over what the same flow would cost in the visual editor. That premium covers the headless render and execution path that the API uses end-to-end. There is no per-request fee on top. The exact credit amount is always reported on the run record — no hidden line items, no surprise add-ons.

Worked examples

A 1-megapixel image typically costs around 4 credits.
A 10-second video from a premium video engine typically costs around 120 credits.
A typical short-form video — script, 3 generated images, 3 image-to-video shots, voiceover and auto-captions — comes out to roughly 210 credits end-to-end.The breakdown varies by which engines you pick for each node. Use the visual editor’s per-node cost display to dial in the exact mix before hitting the API.

Per-request cost controls

integer
Applied per API key. Requests are blocked with 402 monthly_cap_exceeded once the cumulative charges in a UTC calendar month reach the cap. Resets on the first of the following month.
integer
Applied per API key. A run whose estimated cost exceeds this value is rejected at POST /runs time — no credits are deducted.
Set both caps conservatively on any key used by a user-facing integration. A typo in a prompt loop that retries forever is the most common cause of unexpected spend.

Billing lifecycle

Credits are deducted from your balance as the run progresses, and the ledger on your account records every charge so you can audit a single API run down to the individual generation step. If a step fails before any work is billed, the credits for that step are returned to your balance.

Autorefill

Enable autorefill to keep long-running API workloads going without manual top-ups.
1

Settings → API Keys → Autorefill panel

At the top of the API Keys page.
2

Set threshold + topup size

Refill fires when your balance drops below the threshold. Default presets: 400 cr, 850 cr, or 2,200 cr per refill. Pick a topup large enough that you’re not bouncing through it every hour.
3

Attach a card

You’re redirected to our payment processor’s hosted checkout page to collect card details. The card is saved for off-session charges only — Hyperclip never sees your card number.
4

Autorefill fires automatically

When your balance crosses the threshold downward, a refill is queued and the saved card is charged within ~60 seconds. Credits land in your balance shortly after.
If a refill is declined (expired card, insufficient funds, fraud block), autorefill disables itself automatically. The UI shows the error and you must re-attach a card to re-enable. This prevents runaway retry loops against a dead card.
Pick a topup size based on your typical workload: The threshold should be large enough to cover ~1 worst-case run so the refill fires before you hit zero, but small enough that you aren’t sitting on a big unused balance.

Reading your usage

Programmatic usage data is exposed on each run via the credits_charged field:
For aggregate analytics, the UI’s usage dashboard shows per-key monthly totals. Roadmap: a dedicated /v1/usage endpoint.

No hidden fees

  • No per-request fee.
  • No minimum spend. No platform fee.
  • Failed generations (server-side errors, content policy rejections) are fully refunded — you only pay for work that actually ran.
  • Cancelled runs refund any credits not yet committed to in-flight generations; calls that have already started complete (and bill) to their natural end.